Detailed Description
Although the example apparatus described herein includes, among other components, software running on hardware, such apparatus is merely illustrative and should not be considered as limiting. For example, it is contemplated that any or all of the hardware components and software components disclosed may be embodied exclusively in dedicated hardware, exclusively in software, exclusively in firmware or in some combination of hardware, firmware and/or software.
Further, while the following disclosure is presented with respect to an exemplary SPS-based electronic media measurement system, it should be understood that the disclosed apparatus may be readily applied to many other electronic media measurement systems. Thus, while the following describes example apparatus, methods, and articles of manufacture, persons of ordinary skill in the art will readily appreciate that the disclosed examples are not the only way to implement such systems.
In general, the example apparatus, methods, and articles of manufacture described herein may be used to process data that describes a plurality of locations traversed by a responder. In some examples described herein, the data is processed such that the processed data better represents a path of travel along a known route of travel (e.g., street, road, lane, highway, interstate, bridge, sidewalk, pedestrian walkway, railway, tunnel, etc.). In some other examples described herein, the data is processed to express a travel path through a region associated with obstructed signal reception and to mitigate defects present in the data. These defects may include large gaps between locations traversed by the responder, inaccurate location data, and the like.
Further, the example apparatus, methods, and articles of manufacture described herein may be used to associate locations traversed by responders with media sites, thereby scoring media exposure for the media sites. In some examples presented herein, media exposure is scored for a media site if a responder travels within a geometric impact area (geometric impact area) associated with the media site in a direction that facilitates viewing or attraction of the media site. Other examples described herein apply constraints that are to be met before media exposure is scored. Further examples presented herein relate to positions worn by responders where the media site is located in an area where signal reception is impeded and where the media site is a mobile media site.
The example apparatus, methods, and articles of manufacture described herein may be used to reconcile media exposure score data to remove statistical anomalies that do not characterize true media exposure. In some examples, the media exposure credit data is processed such that the credit is within a predetermined amount of an expected value while maintaining the average station travel estimated value. As a result, the examples described herein may be used to improve the accuracy and reach of electronic media metering systems.
Inaccurate or missing data (e.g., in the recorded positioning sequence, or in the media site location information) may adversely affect the accuracy of the media exposure score determined by the media exposure calculation device. To substantially improve the accuracy and reliability of electronic media exposure metering, recorded travel path data may be processed using the example methods and apparatus described herein to overcome the deficiencies discussed above.
Fig. 2 illustrates an example of an SPS enabled device 200 that may be used to implement the monitoring device 110 of fig. 1. To receive and decode signals transmitted by the plurality of SPS satellites 105A-C (i.e., SPS signals), the apparatus 200 includes an SPS signal receiver 205, an SPS signal decoder 210, and an antenna 215. Using any of a variety of techniques, the SPS signal receiver 205 converts a Radio Frequency (RF) analog signal received by the antenna 215 into a digital baseband signal (i.e., a received signal) suitable for processing and/or decoding by the SPS signal decoder 210. For example, the SPS signal receiver 205 may be implemented using a demodulator, a down-converter, a filter, and/or an analog-to-digital (a/D) converter. Using any of a variety of well-known techniques, the SPS signal decoder 210 processes the received signals, if possible (i.e., if no less than a minimum number of SPS satellites 105A-C are available (e.g., in a GPS system, the SPS signal decoder 210 uses received signals from at least 3 or 4 satellites)), to determine the current position of the device 200 (i.e., to make a position fix). SPS signal decoder 210 provides the current geographic position of device 200 (if that position is determined) and the received signals to processor 220. The processor 220 records both the position fix and the received signals (i.e., pseudorange data) in the memory 225. By periodically performing the foregoing method, the recorded data represents the travel path traversed by the responder 102 (FIG. 1).
The example apparatus 200 of fig. 2 also includes an interface 230 that enables the apparatus 200 to communicate with the download server 120 of fig. 1. The device 200 provides the recorded travel path data 305 (i.e., the sequence of position fixes and received signals recorded by the device 200) to a Media Exposure Calculation Device (MECD)300 (discussed below in connection with fig. 3) via the download server 120.
It will be apparent to those of ordinary skill in the art that the processor 220 of fig. 2 may monitor additional data related to the operation, status, etc. of the device 200 and record such additional data in the memory 225. For example, the processor 220 may monitor battery usage, the number of times the device is turned on and off, software errors, and the like.
To improve the stability and reliability of the media exposure scoring decisions made by the MECD 300, the travel path traversed by the respondent 102 is preferably accurate (i.e., reflects the actual location traversed by the respondent 102), follows one or more known routes (e.g., streets, roads, lanes, highways, interstates, bridges, sidewalks, pedestrian walkways, railways, tunnels, etc.), and contains locations that are sufficiently close together. However, as previously mentioned, the sequence of position fixes (i.e., the recorded travel path data 305) recorded by the device 200 may not always meet these requirements.
Fig. 3 is a schematic diagram illustrating an example MECD 300 constructed in accordance with the teachings of the present invention that may be used to implement the example computing device 125 of fig. 1. To post-process the recorded travel path data 305 and media site information (contained in the database 130), the MECD 300 of fig. 3 includes a pre-processor 308. The example preprocessor 308 of fig. 3 includes a travel path processor 310 that operates on recorded travel path data 305 (including both determined geographic locations and received signals (i.e., pseudorange data) recorded by the device 200 and provided via the download server 120) to generate enhanced travel path data 315. In the illustrated example, the recorded travel path data 305 and the enhanced travel path data 315 are stored in one or more memories and/or storage devices implemented as part of the MECD 300. It will be readily apparent to one of ordinary skill in the art that the recorded travel path data 305 and the enhanced travel path data 315 may be implemented in other ways. For example, using a memory or storage device connected with the MECD 300 and configured to communicate with the MECD 300.
The travel path processor 310 processes the recorded travel path data 305 to enhance the integrity and accuracy of the position fix. For example, the travel path processor 310 may use the recorded received SPS signals to derive a position fix (e.g., a location where the device 200 cannot determine a geographic location), increase the accuracy of the position fix determined by the device 200, and so on. The travel path processor 310 may also include additional algorithms to compensate for other known SPS limitations, such as clock drift and multipath signal distortion.
FIG. 4A illustrates an example manner of implementing the example travel path processor 310 of FIG. 3. To process the recorded travel path data 305, the example travel path processor 310 of fig. 3 includes a processing engine 405 for operating on the recorded travel path data 305. For example, the processing engine 405 may be implemented as one or more filters that operate sequentially and/or in parallel on the recorded travel path data 305. In the illustrated example of fig. 4A, the processing engine 405 processes (e.g., applies a set of filters to) a set of data points representing all or part of the travel path transferred by the data transfer unit 415 into the memory 410. The processing engine 405 operates on the set of data points to place intermediate values (e.g., modified and/or additional data points created as the output of the filter and used as input to a subsequent filter), if any, back into the memory 410. The final output data point is placed by the processing engine 405 into the enhanced travel path data 315.
As shown in FIGS. 3 and 4A, and discussed below in connection with the example precise ephemeris filter 442 (FIG. 4B), the example processing engine 405 of FIG. 4A may be able to access data 395 provided by the International Geological Society (IGS) via an Internet connection 390. For example, data 395 includes data accurately describing the positions of SPS satellites 105A-C at known times.
In the example shown in fig. 4A, the memory 410 contains recorded received SPS signals, a position fix determined by the apparatus 200, and a position fix derived by the travel path processor 310. The data stored in memory 410 may be stored using any of a variety of suitable techniques. For example, using object-oriented data storage techniques, using data structure arrays, and the like.
The example processing engine 405 may be implemented using any of a variety of technologies. For example, processing engine 405 may be implemented as software and/or firmware running on a general purpose processing device and/or implemented using hardware as a special purpose processing device (e.g., a digital signal processing device) or any combination of software, firmware, and/or hardware.
It will also be apparent to one of ordinary skill in the art that memory 410 may be implemented using any of a variety of technologies. For example, one or more portions of memory or storage used to implement the recorded travel path data 305, or separate memory, storage, and/or hardware registers directly associated with the travel path processor 310 are used. Further, it will be apparent to those skilled in the art that the data transfer unit 415 may be omitted. For example, the processing engine 405 may be configured to read initial data points directly from the recorded travel path data 305.
Fig. 4B illustrates an example filter sequence that may be used to implement the example processing engine 405 of fig. 4A. In the example shown in FIG. 4B, the filters are implemented using object-oriented programming techniques, thereby increasing flexibility in the number, type, order, configuration, interconnections, and the like of filters.
The example filter sequence shown in fig. 4B begins with a NAV estimation filter 440 that uses the set of position fixes determined by the apparatus 200 to create an initial set of derived position fixes. The precise ephemeris filter 442 acquires precise SPS satellite position data 395 (i.e., ephemeris data 395) from the IGS via the internet 390 and uses the ephemeris data 395 to improve the accuracy of pseudorange data (i.e., received SPS signals) recorded by the apparatus 200 using any of a variety of well-known techniques. For example, the ephemeris filter 442 uses each time scale at each data point in the pseudorange data recorded by the device 200 to interpolate between the known positions of the SPS satellites 105A-C at known times (i.e., the ephemeris data 395) to determine the precise satellite positions at the recorded time scale times. The elevation filter 444 then calculates the angle above the horizon for each SPS satellite 105A-C associated with each pseudorange or located data point from the satellite ephemeris data 395 and using standard orbital geometry principles. To improve the accuracy of the position fix derived from the pseudorange data, the elevation filter 444 discards pseudorange data corresponding to SPS satellites 105A-C that are below the horizon.
Next, a non-simultaneous pseudorange (NSPR) filter 446 determines the location of the missing position fix data point (e.g., representing the location where the device 200 cannot determine a position fix) and derives an additional position fix. In an example, the NSPR filter 446 derives a missing position data point using a set of pseudorange data points centered around the missing position data point and an interpolated clock drift value calculated from pseudorange data associated with the missing position data point and a nearest position data point.
A Receiver Autonomous Integrity Monitoring (RAIM) filter 448 processes the travel path to remove errors caused by multipath distortion. Multipath distortion is caused by the reception of SPS transmit signals reflected off of a plurality of surfaces located between one or more SPS satellites 105A-C and the device 200. Thus, the device 200 will receive multiple versions of the SPS transmitted signal, each having different delay and phase characteristics. In the example where the pseudorange data points comprise signals from four or more SPS satellites, the RAIM filter 448 derives a position fix using various permutations from three of the SPS satellites. Specifically, if four satellites (i.e., #1, #2, #3, #4) are available, four fixes (#1#2#3), (#1#2#4), (#1#3#4), and (#2#3#4) are derived for the following combination of satellites. In another example where the pseudorange data points comprise signals from three SPS satellites (e.g., satellites 105A-C), the RAIM filter 448 derives a position fix using various combinations of the last known positions of the three SPS satellites 105A-C and a fourth SPS satellite (not shown). In both of the foregoing examples, the RAIM filter 448 compares the derived position fixes to each other. If the derived position fix is sufficiently consistent, the position fix is included in the path of travel. Otherwise, multi-path distortion is considered to have occurred and the position fix is removed from the travel path data.
After deriving additional positions or improving the accuracy of existing positions, a street constraint filter 450 (discussed below in conjunction with FIGS. 7A-C, 8A-C, and 9A-C) aligns the various positions included in the travel path to coincide with the centerlines of known travel routes. For example, the street constraint filter 450 modifies (i.e., aligns) the derived position fix to a closest point that coincides with a known travel route (e.g., the centerline of the nearest road, sidewalk, etc.), where the closest point may be determined according to a minimum euclidean distance. However, such modifications may cause the travel path to jump or sway from side to side in an erratic or unwarranted manner (e.g., the travel path moves back and forth between two sidewalks on opposite sides of a street). To alleviate this problem, additional processing may be performed by the street constraint filter 450. The street constraint filter 450 may also process the travel path data to ensure consistency of motion. For example, the street constraint filter 450 may determine whether the travel speed indicates that the responder 102 is in or on a vehicle and, if so, ensure that the travel path is consistent with movement allowed by the immediate environment (e.g., bridge, overpass, underpass, one-way road, etc.).
The gap filter 452 derives additional positions such that the enhanced travel path data 315 is comprised of a series of positions, each position being no more than a predetermined distance (e.g., fifteen feet) from the previous and subsequent positions. The additional position fix is derived using any of a variety of standard geometric or trigonometric techniques that take into account straight and curved travel paths and ensure that the additionally derived position fix is aligned with the centerline of the known travel path. Finally, a National Marine Electronics Association (NMEA) filter 454 outputs the enhanced travel path data 315 using a standard data format (e.g., the well-known NMEA-0183 format).
One of ordinary skill in the art will readily recognize that the number, order, type, configuration, etc. of filters used to implement the processing engine 405 of fig. 4A may be different than the filters shown in fig. 4B. For example, a moving average filter may be used to compute a moving average of the located sequence to smooth the noisy data. Specifically, a running average for each of the last n latitude and the last n longitude may be calculated, where these latitude and longitude are equivalent to the coordinates of the last n position fixes. In another example, a clock drift interpolation filter models drift of a clock used by the device 200 and applies time corrections to the pseudorange data. In yet another example, a dead reckoning filter (dead reckoning filter) estimates a position fix using a previous position fix and an estimated responder direction and speed of travel.
In yet another example, the filters are arranged in two parallel paths. For example, the travel path data 305 is divided into two groups by a data sorting filter. The first group contains data points representing the locations where the respondent 102 appears within a geographic area (e.g., downtown) containing large buildings, and the second group contains data points that are far from downtown. Each data set is then passed through one or more filters, where the filters applied to each data set may be different or the same. Further, data may be exchanged between the two sets of filters (e.g., the two sets of filters may be cross-coupled). The result selection filter is then applied to combine the outputs of the two paths to create the entire travel path for the responder 102.
The travel path processor 310 may operate to further improve the accuracy and reach of the media exposure measurement system as the responder 102 traverses areas where SPS signal reception is obstructed (e.g., subways, tunnels, parking garages, in-buildings, etc.). In an example, the processing engine 405 includes a gap detection filter configured to detect a larger gap in the recorded position fix and to detect that the larger gap begins near a first subway entrance and ends near a second subway entrance (an objective situation in which current SPS techniques cannot be used by devices that are in the underground). If such a large gap is detected, the gap detection filter configures the processing engine 405 to retain the large gap in the enhanced travel path data 315, but the record indicates that the gap is likely to be equivalent to the path of the subway system between the start and end points of the gap. The gap detection filter can similarly detect and record other forms of signal impeded travel processes, such as through a road or pedestrian tunnel, within a parking garage, and the like. The above techniques may be implemented for other types of potentially obstructed signals for determining the location of a responder.
As will be discussed below, the example MECD 300 may use the first and second subway portals to determine possible routes through the subway system connecting the two subway portals. In so doing, MECD 300 may credit media exposure for known media stations located along a subway path that respondents 102 may take or located in a subway train that transports respondents 102 along the subway path. Similarly, if the responder 102 is in front of a media site that is within a otherwise signal-blocked travel path (e.g., within a highway tunnel or pedestrian tunnel, within a building structure, etc.), the media exposure of the media site may be properly integrated.
Returning to fig. 3, the preprocessor 308 of fig. 3 also includes a media site processor 320 in order to process those media sites in the database 130 for which the geo-code location data is missing, or to collate the geo-code location data that exists in the database 130.
Fig. 5 illustrates an example manner of implementing the media site processor 320 of fig. 3. The example media site processor 320 of fig. 5 includes a database access engine 505 that communicates with the database 130 to obtain media site information (i.e., geo-code location data, textual descriptions of media site locations, media site types, etc.) and geo-code location data for known travel routes, reference points, landmarks, etc. The example media site processor 320 also includes an image reader 510, the image reader 510 for accessing images of media sites, which may be stored in the database 130 or available via the internet 390.
In an example, media site processor 320 uses textual descriptions of media site locations along with information describing known travel routes, landmarks, reference points, etc. to derive geo-code location data for the media site. The example media site has a location description indicating that the media site is located west of State street and Main street, the third street. To derive the geo-code location, the example media site processor 320 includes a processing device 515. Knowing the geo-code location of the intersection of the third street and the State street and the intersection of the third street and the Main street, the processing device 515 interpolates between the two known geo-code locations to determine the geo-code location of the example media site.
In another example, the media site processor 320 verifies the accuracy of the geo-code location data by comparing the geo-code location data to a digital representation of an image of the media site (e.g., satellite image, aerial photograph) using image recognition or matching techniques. For example, the media site processor 320 may determine the location of the media site in an aerial photograph of a portion of a city and then compare the location of the media site in the aerial photograph to the geo-code locations of the available media sites.
To determine the location of media sites and other known points (e.g., known travel routes, landmarks, reference points, etc.) in the image, the media site processor 320 includes an image processing engine 520. In an example, the image processing engine 520 determines the location of two intersections and media sites in the image using suitable image recognition and/or matching techniques that are well known. The image processing engine 520 then determines the relative position of the intersection point and the media site. For example, the image processing engine 520 may determine that the media site is located one third between two intersections.
Using the relative location information determined by the image processing engine 520, the processing device 515 can verify the geo-code location data of the media site. For example, in the example introduced above, the processing device 515 uses the known geo-code location data for two location intersections and the determination that the media site is one-third between the two intersections to derive the geo-code location data for the media site. The processing device 515 may then compare the derived geo-code location data for the media site with the geo-code location data for the media site that has been obtained (e.g., contained in the database 130 or determined by the processing device 515 from the textual location description). If the geo-code location data matches, the location of the media site is verified. Otherwise, the location of the media site may be marked for further review and verification.
Processing device 515 and image processing engine 520 may be implemented using any of a variety of techniques. For example, processing device 515 and image processing engine 520 may be implemented using software and/or firmware running on a general purpose processing device and/or a special purpose processing device (e.g., a digital signal processing device), using hardware, or any combination of software, firmware, and/or hardware. Determining the geo-code location from the textual location description and verifying the geo-code location may be done manually.
Returning again to the example illustrated in fig. 3, the processed media site location data 325 is stored within one or more memories and/or storage devices implemented as part of the MECD 300. However, the processed media site location data 325 may be otherwise implemented. For example, using a memory or storage device connected and configured to communicate with MECD 300, within database 130, and the like.
Fig. 6A, 6B, and 6C illustrate flowcharts representative of example machine readable instructions that may be executed by a processor (e.g., one of the processors 2305A-C of fig. 23) to implement the example preprocessor 308 of fig. 3, the travel path processor 310 of fig. 3, and the media site processor 320, respectively. The machine readable instructions of the example preprocessor 308, the example travel path processor 310, and/or the example media site processor 320 of fig. 6A-C may be executed by a processor, a controller, and/or any other suitable processing device. For example, the machine readable instructions of the example preprocessor 308, the example travel path processor 310, and/or the example media site processor 320 of fig. 6A-C may be embodied as coded instructions stored on a tangible medium, such as a flash memory or Random Access Memory (RAM) associated with the processors 2305A-C shown in the example processor platform 2300 discussed below in connection with fig. 23. Alternatively, some or all of the example pre-processor 308, the example travel path processor 310, and/or the example media site processor 320 of fig. 6A-6C may be implemented using Application Specific Integrated Circuits (ASICs), Programmable Logic Devices (PLDs), Field Programmable Logic Devices (FPLDs), discrete logic, hardware, software, and/or firmware. Some or all of the example machine readable instructions of the example preprocessor 308, the example travel path processor 310, and/or the example media site processor 320 of fig. 6A-6C may be implemented manually or as a combination of any of the foregoing techniques. 6A-6C are described with reference to the flow diagrams of FIGS. 6A-6C, persons of ordinary skill in the art will readily appreciate that many other methods of implementing the example preprocessor 308, the example travel path processor 310, and/or the example media site processor 320 may be employed. For example, the order of execution of the blocks may be changed, and/or some of the blocks described may be changed, eliminated, or combined.
The example machine readable instructions of fig. 6A begin with the preprocessor 308 processing various media sites in the database 130 using the example machine readable instructions of fig. 6B (block 602). Although the pre-processor 308 is described as processing all of the media sites in the database 130, the pre-processor 308 may alternatively process only a portion of the media sites in the database 130. For example, the preprocessor 308 may only process media sites that are associated with certain demographic or market regions (e.g., cities or metropolitan areas).
In the example machine readable instructions of fig. 6A, the preprocessor 308 then reads a configuration file (block 604) that specifies which filters and filter configuration(s) to implement by the travel path processor 310. In one example, the configuration file is an XML file that identifies the type, order, sequence, configuration, interconnections, and number of filters. However, other types and/or numbers of filters may be used instead.
The preprocessor 308 next processes the travel path data corresponding to each responder (block 606) (block 608) using the example machine readable instructions of FIG. 6C. If all of the respondent's corresponding travel path data has been processed (block 610), the preprocessor 308 ends execution of the example machine readable instructions of FIG. 6A. Otherwise, the preprocessor 308 returns to block 606 to process the travel path corresponding to the next responder.
The example machine readable instructions of fig. 6B begin with the media site processor 320 processing various selected media sites from the database 130 (block 630). For each media site, the processor 320 determines whether the geo-code location data is available (block 632). If the geo-code location data corresponding to the media site is not available (block 632), the media site processor 320 determines if a textual location description for the media site is available (block 634). If a textual location description is not available (block 634), the media site processor 320 marks the media site for further processing (e.g., error processing) (block 635). Returning to block 634, if a textual location description is available, the media site processor 320 derives (as previously described) geo-code location data from the textual location description (block 636).
Returning to block 632, if the geo-code location data is available, the media site processor 320 determines whether an image containing the media site is available (block 640). In the example of fig. 6B, the media site processor 320 determines whether the image can be obtained by examining the images in the database 130 and/or via an internet site. If an image is available (block 640), the media site processor 320 reads the image (block 642); determining the location of the media site and nearby landmarks, known travel routes, reference points, etc. using the known geo-code location data; and determines the geo-code location data corresponding to the media site from the image (block 644). The media site processor 320 then determines whether the geo-code location data determined from the image sufficiently matches the geo-code location data present in the database 130 or derived from the textual location description (block 646). If the geo-code location data sufficiently matches (block 646), the media site processor 320 stores the geo-code location data in the media site location data 325 (block 650). Otherwise, if the geo-code location data does not sufficiently match (block 646), the media site processor 320 marks the media site for further processing (e.g., error processing) (block 635).
If all of the selected media sites have been processed (block 652), the media site processor 320 ends execution of the example machine readable instructions of FIG. 6B. Otherwise, if all sites have not been processed (block 652), the media site processor 320 returns to block 630 to process the next media site.
The example machine readable instructions of fig. 6C begin with the travel path processor 310 operating on the various filters specified in the filter profile (discussed above) (block 660). The travel path processor 310 then operates one of these filters (block 662). If all filters have been operated on (block 664), the media site processor 320 ends execution of the example machine readable instructions of FIG. 6C. Otherwise, if not all filters have been operated on (block 664), the travel path processor 310 returns to block 660 to operate the next filter.
Returning to the street constraint filter 450 of fig. 4B, the various derived (or determined) locations within the travel path are aligned (i.e., modified, manipulated, etc.) to coincide with the centerline of the known travel path such that the resulting enhanced travel path data 315 represents a coherent and reasonable travel path along the known travel path. The street constraint filter 450 determines the appropriate and most likely position location based on past and future travel processes. In an example, implementation of the street constraint filter 450 uses Artificial Intelligence (AI) algorithms and techniques (with appropriately chosen subtractive and weighted values) for various travel path manipulations. For example, individual positions may be mapped to a plurality of points corresponding to nearby known travel routes to create a Bayesian (Bayesian) tree representing a plurality of possible travel paths connecting the mapped positions. Values may then be added to each point (e.g., based on the euclidean distance from the actual location to that point). The cost associated with each path is determined by summing the values corresponding to the various mapping points that make up the path, and the path with the lowest cost is selected.
In the example of fig. 3 and 4B, the travel path processor 310 may use geocoded data that indicates the location of a known travel route. In addition, the travel path processor 310 may use a street map file that defines the geographic or demographic area for which the street constraint filter 450 operates. Thus, the portion of the travel path that traverses or traverses the region will be processed by the street constraint filter 450. In the example of fig. 3 and 4B, the street map file is a configurable XML file that defines a simple bounding rectangle defined by four longitude and latitude pairs. The travel path processor 310 uses the bounding rectangle to determine the segments (e.g., 50 feet long) of each known travel route that fall within the area. The travel path processor 310 performs an operation of constraining the locations to align them with the centerline of one of the road segments falling within the area.
FIG. 7A illustrates an example travel path portion including 20 derived positions (represented by circles 1-20). Within the example street constraint filter 450, a travel segment is a sequential set of consecutive data associated with a certain known travel route. For example, in fig. 7A, Pine street has three travel segments associated with it: (1, 2, 3, 4, 5), (13, 14, 15, 16) and (19, 20).
A deterministic path is constructed by forcing each position fix to be associated with only one segment of a known travel route. Fig. 7B illustrates an example deterministic path constructed from the example travel path shown in fig. 7A, where each node in the example deterministic path is equivalent to one travel segment. If the street constraint filter 450 only considers deterministic paths, then it is likely that the known travel route that a point appears to be closest to is not actually the known travel route along which the responder 102 traveled. For example, in the example of fig. 7A, the location 17 may be associated with a second street or Pine.
Given the foregoing reliance on deterministic paths, the example street constraint filter 450 constructs a decision tree that includes a plurality of mappings from fixes to potentially known travel routes. Thus, a decision tree is made up of the possible travel paths corresponding to these positions, where the complexity of the decision tree depends on the amount of ambiguity in the position (e.g., the number or percentage of ambiguity points). Each node in the decision tree represents a travel segment (i.e., a candidate segment) of a candidate travel path. FIG. 7C illustrates an example decision tree containing two branches constructed from the example travel path data shown in FIG. 7A. The example decision tree of fig. 7C is relatively small because the travel path data has a relatively low amount of ambiguity.
By constructing a decision tree, the street constraint filter 450 may employ fuzzy logic by setting a set of rules to determine the likelihood that each of the candidate travel paths comprising the decision tree is the actual travel path taken by the responder 102. Specifically, each candidate travel path is assigned a score, and the candidate travel path with the highest score is the most likely travel path to be taken by the responder 102.
In the example street constraint filter 450, the current location is found to be most affected by the nearest neighbor. For example, in the example of FIG. 7A, whether fix 17 should be on Pine or on the second street is primarily affected by fixes 16 and 18. Thus, the example street constraint filter 450 uses a predictor-modifier algorithm. For example, to determine the best known travel route to which a position fix is mapped, the example street constraint filter 450 iterates through the travel path data until a decision tree of predetermined depth (e.g., four) is constructed. The example street constraint filter 450 then determines a score for each branch in the finite depth tree and picks the branch with the highest score. After making a decision on one position fix (or candidate segment), the example street constraint filter 450 repeats this process for the next position fix (or candidate segment).
Various methods (i.e., criteria) may be used to score the individual branches of the finite depth decision tree. For example, the proximity of the position fix to the candidate segment, the apparent alignment of the position fix with respect to the candidate segment, etc. FIG. 8A illustrates additional example positioning. Example criteria are based on data moments, such as data moments taken with respect to candidate road segments. Fig. 8B and 8C illustrate two moments of the example location of fig. 8A taken with respect to a first street and a second street, respectively. Candidate segments with smaller average distances or moments are rated higher than candidate segments with higher average distances or moments. In the example street constraint filter 450, the data moments are used as initial scores assigned to candidate road segments (i.e., nodes of the decision tree).
Another example criterion is a scalar product (dot product), which measures how well a candidate road segment is aligned with a corresponding position. The scalar product of the candidate segment and the position fix determines the angle between the position fix and the candidate segment. In this example, if the angle is close to 0 or 180 degrees, the travel segment (i.e., decision tree node) is assigned a higher rank (i.e., is given a bonus), and if the angle is close to 90 or 270 degrees, the travel segment is decremented.
Yet another example criterion utilizes context analysis (contextqualanalysis) based on candidate road segments. For example, consider a candidate segment s [ n ]. FIG. 9A lists some example contextual analysis bonus points awarded to candidate road segments s [ n ]. Specifically, if s [ n ] has more than five consecutive points (i.e., positions), then the candidate segment s [ n ] is awarded a bonus of 40% (i.e., its score is increased by 40%). If the score of the previous candidate segment s [ n-1] is greater than a predetermined amount (e.g., 60), a bonus of 10% of the candidate segment s [ n ] is awarded.
FIGS. 9B-G illustrate example candidate path configurations each resulting in a 15% venation analysis score. For example, as shown in FIG. 9C, if the candidate segment s [ n ] and s [ n +1] are not connected, a 15% reduction is made to the candidate segment s [ n ].
Returning to fig. 3, to determine whether the responder 102 is exposed at the media site 115, the MECD 300 of fig. 3 includes a transit processor 328. The traversal processor 328 in the example shown in fig. 3 uses the enhanced travel path data 315, the media site location data 325, and a set of constraints to determine whether the responder 102 has traversed the media site 115 (fig. 1) in a way that has a chance to see the media site 115. For the media site 115 to be exposed to the bonus media in the example shown in fig. 3, the responder 102 must traverse what influence zone is associated with the media site 115 in a direction that facilitates viewing of the media site 115.
An example zone of influence 1010 associated with a media site 115 is illustrated in fig. 10. In the example of fig. 10, the media sites 115 are oriented 315 degrees north-west. The example area of influence 1010 is a geometric area constructed as a portion of a circle 1005 having a radius equal to the maximum distance from the media site 115 at which the responder 102 may view the media site 115, where lines 1020 and 1025 represent the maximum angle of the range at which the responder 102 may view the media site 115. The maximum viewable distance varies depending on the type and make-up of the media site 115. In the example shown in fig. 10, the maximum viewable distance is directly related to the site type and depends on the site type (e.g., the size of the media site). For example, media stations located on the side of a bus shelter are typically visible from an appearance of approximately 210 feet, and 20 foot by 60 foot electronic billboards are typically visible from an appearance of 1400 feet. Further, the maximum viewable distance associated with a media site may vary depending on the location of the media site, e.g., a media site 25 feet away from the ground may be viewed at a greater distance. Additionally, the maximum viewable distance may be related to the particular font, color, etc. employed by the media site. Although the example area of influence 1010 is depicted as a portion of a circle, the area of influence 1010 may be configured as a geometric area of a different shape, such as a rectangular area.
In the example shown in fig. 10, the maximum viewable angle is 140 degrees centered on vector 1013, which vector 1013 is at an orientation along the direction that media site 115 is facing (i.e., 315 degrees). Thus, line 1020 corresponds to 25 degrees (i.e., (317+ 70)% 360 degrees) and line 1025 corresponds to 245 degrees (i.e., (315-70)% 360 degrees), where the symbol% represents the modulo operator. However, for some media sites, the maximum viewable angle is 180 degrees. This is equivalent to line 1020 coinciding with the 45 degree line and line 1025 coinciding with the 225 degree line. For a media site 115 to credit media exposure, the responder 102 must either traverse the zone of influence 1010 associated with the media site 115 in a favorable direction or traverse within the zone of influence 1010.
The direction of travel that is helpful to see the media site 115 or that is attracted to the media site 115 depends on the direction the media site 115 is facing and the maximum responder visibility angle. The maximum responder visibility angle is the range of angles at which the responder 102 can see the media site 115 without turning around or is attracted to the media site 115. For example, 65 degrees is the typical range of angles that most responders can see when sitting in a car and limited by the car windshield, while 50 degrees is the range of 90% human visual realization that studies have shown. In particular, for X-degree oriented media sites 115 and Y-degree maximum responder visibility angles, advantageous directions of travel include directions of travel between [ (X- (Y/2) + 180)% 360] degrees and [ (X + (Y/2) + 180)% 360] degrees, where the symbol% represents the modulo operator. Thus, with a maximum responder visibility angle of 140 degrees, the favored direction of travel in the example shown in FIG. 10 is between 65 and 205 degrees.
FIG. 11 illustrates a straight travel path 1105 passing at 90 degrees (straight east) within the example area of influence 1010 of FIG. 10. The portion of the straight path 1105 represented by arrow 1110 does not result in media exposure scoring of the media site 115 because the responder 102 is outside the area of influence 1010 (the shaded region). The portion of the travel path 1105 represented by the multiple bold arrows 1115 through 1120 may result in media exposure scoring of the media site 115 because the responder 102 is traveling within the influence area 1010 in a direction conducive to viewing the media site (i.e., 65 degrees < 90 degrees < 205 degrees). The remainder of the travel path 1105 represented by arrow 1125 does not result in media exposure credits being made to the media site 115 because the responder 102 has left the influence area 1010.
FIG. 12 illustrates a curved travel path 1205 traversing through the example area of influence 1010 of FIG. 10 and traversing within the example area of influence 1010. The portion of the travel path 1205 that travels along the favored travel direction within the area of influence 1010 (shaded area) is represented as a bold arrow.
Referring to fig. 3, to calculate the impact zone 1010 associated with a media site 115, the channel processor 328 includes an impact zone calculation device 330, the device 330 calculating the impact zone 1010 based on the maximum viewable distance and the maximum viewable angle associated with the media site 115 recorded in the media site location data 325. To determine whether the respondent 102 has crossed the area of influence 1010 calculated by the area of influence calculation device 330, the travel processor 328 includes a location comparison device 335 that compares the locations in the enhanced travel path data 315 to determine whether they fall within the area of influence 1010. The area of influence computing device 330 also computes a range of favorable viewing directions for the media site 115 based on the direction the media site 115 is facing and the maximum responder visibility angle. The range of favored viewing directions for the media site 115 may be calculated for the maximum responder viewing angle applicable to all respondents. Alternatively, the range of favored viewing directions for the media site 115 may be calculated for each responder, thereby facilitating a maximum responder viewing angle for each responder.
To determine the direction of travel, the travel processor 328 includes a direction of travel calculation device 340 that calculates a direction of travel for a location that falls within the area of influence 1010. The position comparison means 335 provides the position fix falling within the area of influence 1010 to the direction of travel calculation means 340. The direction of travel associated with a location is determined using at least one other location and using standard theory of location. For example by constructing a vector connecting the position with the next position and determining the direction associated with the constructed vector.
The travel processor 328 also includes a direction comparator 345 that compares the travel direction corresponding to a location falling within the influence zone 1010 with the range of favorable travel directions calculated by the influence zone calculation device 330.
Each media site is treated separately, even if the media sites are located close to each other, and each media site is associated with an area of influence and a favorable direction of travel. For example, in the case of two bulletin boards placed back-to-back and perpendicular to a highway, one bulletin board gets integrated as viewed by a responder traveling in one direction, while the other gets integrated as viewed by a responder traveling in the opposite direction.
As will be discussed in more detail below, additional constraints (e.g., station illumination, exit and re-entry into the area of influence, etc.) may be applied such that the media station 115 may not be integrated by exposing the responder 102 located on a position even if this position satisfies the area of influence and the favorable direction of travel constraint, i.e., is located within the area of influence and the responder 102 is moving along and/or facing the favorable direction of travel. For example, if the responder 102 passes the media site 115 without natural light and the media site 115 is not illuminated, the media site 115 should not be credited with exposure. To apply additional constraints, the path processor 328 includes a constraint processor 350. The individual exposures integrated for the media site 115 are recorded in the database 130 by the constraint processor 350.
The media site information present in the database 130 specifies whether the media site 115 is illuminated and, if so, the number of hours of illumination. For example, some media sites are not illuminated, and thus they can integrate media exposure only in the presence of natural light. For example, in chicago, illinois, daylight hours are approximately 12 hours (6 am to 8 pm) in the winter (4 to 9 months). In addition, sunshine hours may be determined daily using a meter capable of measuring natural lighting conditions or using weather data reporting sunrise and sunset times. For a media site 115 that is illuminated, the media site 115 may score the media exposure similarly during times of day and periods when the media site 115 is illuminated. Additionally or alternatively, the scoring of the media site 115 for viewing by respondents with diseases that cause or result in vision degradation (e.g., night blindness) may be adjusted to account for this or other vision condition known to affect the ability of respondents to view media from far distances.
When the responder 102 has multiple consecutive positions that fall within the influence area 1010, the media site 115 only integrates one exposure. Specifically, if 150 or more consecutive positions are located within 50 feet (except for 5 at most), then the sequence of positions is considered a series and only one exposure is integrated. If more than 5 points in the continuous list are outside 50 feet, the exposures are integrated multiple times.
To cope with situations where respondents 102 enter and exit the area of influence 1010 multiple times very close to the edge of the area of influence 1010, additional constraints are applied. If the responder 102 leaves the influence zone 1010 and then re-enters the influence zone 1010, the media site 115 is not integrated for another exposure unless the length of time the responder 102 leaves the influence zone 1010 reaches a minimum period of time. In the example shown, the minimum time period is 10 minutes. However, any other time period may be used instead.
Typically, the media site 115 is located along one highway (i.e., a primary road) while the responder 102 is traveling along a second road (i.e., a secondary road) in a favorable direction and enters the area of influence 1010 of the media site 115. The constraints to deal with this situation may differ depending on the location of the media site 115. In the illustrated example, the media site 115 is only credited with a media exposure if the responder 102 is traveling on a primary road or on a secondary road in a predetermined list, where the predetermined list of secondary roads includes secondary roads on which the media site 115 is viewable and to which the owner of the media site 115 desires to join the list. In addition, the media sites 115 may be categorized according to whether the media sites 115 are surrounded by tall buildings (e.g., a commercial district) that may affect the viewing of the media sites 115. For example, if the media site 115 is surrounded by tall buildings, the media site 115 will only score media exposures if the responder 102 is traveling on a primary road or if the media site 115 is on a rooftop (and thus visible from a secondary road). If the media site 115 is not surrounded by tall buildings, the media site 115 is scored for media exposure regardless of whether the responder 102 is traveling on the primary road or the secondary road.
The example methods discussed above may be used to determine media exposure for mobile media sites (e.g., media on the side of a vehicle cabin). For example, most buses have four possible advertising words per vehicle (i.e., media sites) -one on each side of the bus (i.e., front, rear, passenger boarding and driver seat). FIG. 13 is a schematic diagram illustrating four zones of influence 1310, 1315, 1320, and 1325 created around a bus 1305. If other media targeting arrangements are employed (e.g., a single word is spread throughout the bus), the method outlined below may be modified in ways that will be apparent to those of ordinary skill in the art.
In the example shown in FIG. 13, the responder 102 may only see one of the four advertising words at any given time. Thus, the bus 1305 is divided into four quadrants, each quadrant corresponding to one side of the bus 1305. To illustrate this, we can imagine oneself standing on the roof of the bus 1305, facing the direction of travel of the bus 1305. If the letter "X" is drawn on the roof, each of the four advertising words will result in one of the four quadrants formed by the "X". In the example shown in FIG. 13, the four quadrants correspond to the four zones of influence 1310, 1315, 1320, and 1325, respectively. One of ordinary skill in the art will readily recognize that other impact zone shapes may be used. For example, if the ad words have a maximum viewable angle of 180 degrees, the impact zones may overlap and the responder 102 may see two ad words or media displays at the same time. The favorable direction of travel for each ad word for the bus 1305 may be calculated in a manner similar or identical to that used for fixed media sites.
For fixed location media sites, the exact location of the media site 115 is determined from the geo-code location data obtained in the media site location data 325. However, in the case of the bus 1305, the coordinates of the bus 1305 may be constantly changing because the bus 1305 may stop and go between stations. Each bus route has a predefined planned parking position and time, even if the bus is moving nearly continuously. In this way, the travel route of the bus 1305 may be simulated using the planned parking position and time specified in advance and the bus route assigned to the bus 1305. The geo-code location data for each planned parking location may be readily obtained using any of a variety of known techniques. According to the use of these known fixed locations, the affected areas 1310, 1315, 1320, and 1325 around the bus 1305 parked at the bus stop may be treated like four fixed outdoor media stations (e.g., bulletin boards or bus shelters) for a particular period of time and the previously discussed method of determining media exposure may be employed. Specifically, any one pass of responder 102 through impact zones 1310, 1315, 1320, and 1325 in a favorable direction during the time that bus 1305 is parked at a bus stop will be recorded as a media exposure.
At the end of each planned bus stop, the zones of influence 1310, 1315, 1320, and 1325 will move in the same direction with the bus 1305. As the bus 1305 moves forward, a new zone of influence is created along a simulated travel path representing another set of fixed media sites having zones of influence adjacent to the zones of influence 1310, 1315, 1320, and 1325 associated with the bus stop. These new influence areas are the same size as the original set of influence areas 1310, 1315, 1320, and 1325. Any one of these new impact zones traversed by the responder 102 within a predetermined time window determined from the estimated location of the bus 1305 at a given time may result in the scoring of the relevant advertising words for viewing. In other words, the influence zone may be created at each bus stop using a bus schedule. Between any two known bus stops, a virtual bus stop (and associated zone of influence) may be created such that the zone of influence is contiguous between the known bus stops and spreads out along the simulated travel path of the bus 1305. The start and end times for each virtual station may be calculated by interpolating the known times between planned stops. This approach allows for the possibility of media exposure integration at any time while the bus 1305 is on the road as planned and at any location throughout the bus route. In addition, the impact zone may be considered as a continuously moving impact zone, rather than being quantized into segments.
As shown in fig. 13, the distance at which the headlight signal can be seen (in front of the bus 1305) is within approximately 50 feet in front of the bus 1305, the bus is about 24 feet long, and the distance at which the taillight signal can be seen is within about 75 feet behind the bus. Thus, in the example of FIG. 13, a total street distance of 150 feet is used as the length of the road segment.
Fig. 14 illustrates an example scenario involving three city blocks, including bus stops (stops #14 and #15 on the route) at the first and third blocks in this figure and two virtual stops 1405 and 1410 located between the two known stops, four non-overlapping zones of influence for the front of the bus 1305, four zones of influence for the driver side of the bus 1305, and so on.
The same constraints associated with lighting discussed above with respect to fixed media sites can be applied to mobile media sites. Additionally, the bus 1305 may be equipped with an SPS device to record the actual stop location, rather than derive the bus location from an associated bus schedule. In this case, the actual bus location data may be used to derive the area of influence and identify a favorable direction of travel for any location and/or locations (depending on the granularity of the data desired) at which the bus 1305 is located at any time.
Fig. 15, 16, and 17 illustrate flowcharts representative of example transit processors 328 of fig. 3 that may be executed by a processor (e.g., one of the processors 2305A-C of fig. 23) to implement. The machine-readable instructions of the example of fig. 15-17 and/or fig. 3 traveling through processor 328 may be executed by a processor, a controller, and/or any other suitable processing device. For example, the machine-readable instructions of the example travel processor 328 of fig. 15-17 and/or fig. 3 may be embodied in coded instructions stored in a tangible medium (flash memory) or Random Access Memory (RAM) associated with the processors 2305A-C shown in the example processor platform 2300 and discussed below in connection with fig. 23. Additionally, some or all of the example machine readable instructions of the example processors 328 of fig. 15-17 and/or 3 may be implemented using Application Specific Integrated Circuits (ASICs), Programmable Logic Devices (PLDs), Field Programmable Logic Devices (FPLDs), discrete logic, hardware, software, and/or the like. Also, some or all of the example machine readable instructions of the example processors 328 of fig. 15-17 and/or fig. 3 may be implemented manually or as a combination of any of the foregoing techniques. Further, although the example machine readable instructions of FIGS. 15-17 are described with reference to the flowcharts of FIGS. 15-17, persons of ordinary skill in the art will readily appreciate that many other methods of implementing the example traversal processor 328 of FIG. 3 may be used. For example, the order of execution of the blocks may be changed, and/or some of the blocks described may be changed, eliminated, or combined.
The example via processor 328 executes the example machine readable instructions of fig. 15 beginning at block 1505, and at block 1505, processes all media sites. The traversal processor 328 calculates an area of influence 1010 associated with the media site 115 based on the maximum viewing distance, the facing direction, and the maximum viewing angle of the media site 115 (block 1510). The travel processor 328 then calculates a range of favored travel directions (block 1515). As previously discussed, the range of favorable directions of travel may correspond to a common maximum responder visibility angle or be associated with a maximum responder visibility angle for each responder. The traversal processor 328 proceeds to block 1520 where all the fixes associated with all the respondents are processed. For one location, the navigation processor 328 compares the location to the area of influence 1010 (block 1525). If the position fix falls within the influence zone 1010 (block 1525), the route processor 328 compares the direction of travel of the responder 102 at the position fix to a range of favorable directions of travel associated with the media site 115 (block 1530). If the direction of travel of the responder 102 is favorable (block 1530), the travel processor 328 applies additional applicable constraints by executing the example machine readable instructions of FIG. 17 (block 1532).
Returning to block 1525, if the position fix does not fall within the area of influence 1010, then the via processor 328 determines whether all position fixes have been processed (block 1535). If there are more locations not processed (block 1535), then the travel path processor 328 returns to block 1520 to process the next location. Otherwise, if all the positions have been processed (block 1535), the path processor 328 determines whether all the media sites have been processed (block 1540). If all of the media sites have been processed (block 1540), the travel processor ends execution of the example machine readable instructions of FIG. 15. Otherwise, if there are more media sites that have not been processed (block 1540), the path processor 328 returns to block 1505 to process the next media site.
To determine the media exposure of a mobile media site (e.g., bus 1305), block 1505 of FIG. 15 includes a plurality of temporary impact zones associated with the mobile media site. Further, the determination of whether the responder 102 is within the temporary impact zones (block 1525) is affected by the validity period corresponding to each temporary impact zone.
For portions of the travel path associated with blocked signal reception, an additional decision block may be added before block 1525 in the example machine readable instructions of FIG. 15. The additional decision block detects gaps in positioning related to obstructed signal reception (noted in the enhanced travel path data by the pre-processor 308 of fig. 3). If a gap is detected, the travel path processor 328 determines a likely path for the portion of the travel path associated with the obstructed signal reception and integrates the media exposure for the media sites along the likely path.
A more computationally efficient implementation of the example machine readable instructions shown in FIG. 15 is illustrated in another alternative example machine readable instructions of FIG. 16. The example machine readable instructions of FIG. 16 take advantage of the fact that aligning the responder travel path to become along a known travel path and determining whether a segment of the known path requires less computation to fall within a square than a circle.
The processor 328 executes the alternative example machine readable instructions of fig. 16 beginning at block 1605, where all media sites are to be processed at block 1605. The traversal processor 328 calculates a square area centered on the media site 115 such that each side of the square area has a length approximately equal to twice the maximum viewing distance of the media site 115 (block 1610). However, other lengths or distances may be used instead. Further, the area may be another polygonal shape than rectangular or square, such that it roughly corresponds to the size of the area of influence 1010 associated with the media site 115. The travel path processor 328 also determines a list of road segments that are associated with known travel paths that fall within the square region (block 1615).
The traversal processor 328 then calculates an area of influence 1010 associated with the media site 115 based on the maximum viewing distance, facing direction, and maximum viewing angle of the media site 115 (block 1620). The travel direction processor 328 next calculates a favorable travel direction (block 1625). As previously discussed, the favored-travel direction may correspond to a common maximum responder visibility angle or be associated with a maximum responder visibility angle for each responder.
The traversal processor 328 proceeds to block 1630 where all locations associated with all respondents are processed. For one location, the travel path processor 328 compares the location (which has been aligned with a segment of the known travel path) to a list of road segments for which the known travel path is contained within a square region (block 1635). If the location falls on one of the road segments (block 1635), the travel processor 328 compares the location to the area of influence 1010 (block 1640). If the location falls within the area of influence 1010 (block 1640), the travel processor 328 compares the direction of travel of the responder 102 at the location to the favored direction of travel associated with the media site 115 (block 1645). If the direction of travel of the responder 102 is favorable (block 1645), the travel processor 328 applies additional applicable constraints using the example machine readable instructions of FIG. 17 (block 1650). Otherwise, if the direction of travel of the responder 102 is not favorable (block 1645), then the travel processor 328 proceeds to block 1655.
If additional positions have not been processed (block 1655), then the travel processor 328 returns to block 1630 to process the next position. Otherwise, if all the positions have been processed (block 1655), the traversal processor 328 determines whether all the media sites have been processed (block 1660). If all of the media sites have been processed (block 1660), the processor ends execution of the example machine readable instructions of FIG. 16. Otherwise, if there are more media sites that have not been processed (block 1660), then the travel processor 328 returns to block 1605 to process the next media site.
The travel processor 328 begins execution of the example machine readable instructions of fig. 17 by comparing the time associated with a position fix to the sunshine time associated with the media site 115 (block 1705). If the time falls outside of the sunshine hours of the media site 115 (block 1705), the travel processor 328 determines if the time falls within the lighting times, if any, of the media site 115 (block 1707). If the travel processor 328 determines that the site is not illuminated at the time associated with the position fix (block 1707), the travel processor 328 ends execution of the example machine readable instructions of FIG. 17 because the media site 115 is unlikely to be seen by the responder 102 without a sunshine condition (block 1705) or site illumination (block 1707).
If a sunshine condition exists (1705) or the media site 115 is artificially illuminated for a time associated with a position fix (block 1707), the travel path processor 328 determines whether the travel path includes a plurality of position fixes located within the area of influence 1010 (block 1710). If multiple locations are within the influence zone 1010 (block 1710), the traversal processor 328 determines whether the responder 102 has been outside the influence zone 1010 for at least 10 minutes since the responder 102 was last within the influence zone 1010 (block 1715). If the responder 102 has not been outside the influence zone 1010 for at least 10 minutes (block 1715), the traversal processor 328 ends execution of the example machine-readable instructions of FIG. 17. It will be apparent to one of ordinary skill in the art that any other time period may be used instead of ten minutes, and preferably the length of the time period is related to the likelihood that a person outside the area of influence 1010 for a specified length of time will focus their viewing attention on the media site 115 when re-entering the area of influence 1010, thereby making the individual one-shot exposure scores reasonable.
If there are no multiple positions within the area of influence 1010 (block 1710) or if the responder 102 has walked out of the area of influence 1010 for more than 10 minutes (block 1715), the travel processor 328 determines whether the positions are along a primary road associated with the media site 115 (block 1720). If these locations are along the primary road (block 1720), the travel processor 328 integrates the exposure for the media site 115 (block 1725) and ends execution of the example machine readable instructions of FIG. 17.
Returning to block 1720, if the locations are not along a primary road, the travel processor 328 determines whether the locations are along a predetermined secondary road where the media site 115 can be seen (block 1730). If the positions are along the predetermined secondary road where the media site 115 can be seen (block 1730), the travel processor 328 credits the exposure for the media site 115 (block 1725) and ends execution of the example machine readable instructions of FIG. 17. Otherwise, if the locations are not along a secondary road where the media site 115 is visible (block 1730), the traversal processor 328 ends execution of the example machine readable instructions of fig. 17 without integrating the exposure of the responder 102 with the media site 115. Additionally, the example machine readable instructions of fig. 17 may skip decision block 1730 and only integrate exposure for the media site 115 for travel on the primary road.
Statistical sampling errors are inevitable for any media exposure evaluation system. For example, when the demographic population of the plurality of respondents 102 does not completely match the demographic distribution of the promotional area; when the number of respondents 102 is not large enough to ensure that all media sites 115 are traversed; when a responder 102 lives very close to several media sites 115 and, therefore, the exposure scores for those media sites 115 are too high; etc., may cause errors. In addition, the population of respondents 102 may not give a complete set of demographic data.
Returning to fig. 3, to improve the statistical accuracy and/or representativeness of the media exposure scores (i.e., the travel data) determined by the travel processor 328 and stored in the database 130, the MECD 300 includes a statistical processing device 397. The statistical processing device 397 employs statistical analysis algorithms to reconcile the travel data to be more representative of a larger population of responders, to estimate missing demographics, and/or to create coverage and frequency values representative of the effects of media sites. However, it will be apparent to those skilled in the art that it is important that any statistical processing performed by the statistical processing means 397 ensure that the resulting travel data (and any resulting impact ranges and frequency values therefrom) remain a non-biased media site exposure assessment result in each metered market or area.
Fig. 18 illustrates an example manner of implementing the statistics processing apparatus 397 of fig. 3. To coordinate the travel data, the statistical processing device 397 includes a data coordination processor 1805. The coordination processor 1805 seeks to eliminate extreme features of the traversal data (including stations with zero or abnormally high traversal data) and to smooth the traversal data while keeping the total average media site traversal estimates for each media site owner and media site type constant.
In the outdoor media market, additional alternative media site travel estimates based on car traffic surveys are available (e.g., TAB real-time traffic per Day (DEC) travel estimates). The vehicle traffic dependent survey differs from the example electronic media site exposure system of fig. 1 in definition, data method, and timing. However, statistical analysis and comparison of results show that for most media sites, the results from these two techniques are typically within +/-20% of each other.
In an example, the coordination processor 1805 reforms the travel data in a round-robin fashion. In each cycle, the coordination processor 1805 constrains the travel data to be within +/-20% of the TAB DEC travel data and scales the constrained travel data to restore an average travel data value (i.e., constrains the travel data to keep the average constant). Specifically, to scale the constrained travel data, the data coordination processor 1805 calculates a current average, divides the current average by the original average to determine a scaling factor, and multiplies the constrained travel data by the scaling factor. In this example, it is most important to keep the average amount of menstruation constant, so if both constraints cannot be met simultaneously, the percentage constraint is relaxed. For media stations without DEC TAB data, a simple weighted average is performed by the statistical processing means 397 to smooth the traversing data. The statistical processing means 397 ends the loop when the maximum number of loops occurs or convergence is detected. In this example, convergence is determined by monitoring the percentage error between the current average amount of travel and the original average amount of travel. Convergence occurs when the percentage falls below a predetermined limit. Other techniques for data coordination may be used in place of or in addition to the techniques described above. For example, coordination with other media site exposure survey data, use of different target percentages, use of different convergence criteria, and the like may be used. Further, the travel data may be adapted by the statistical processing device 397 to satisfy any number of constraints (including a single constraint).
FIG. 19 illustrates example travel data before, during, and after coordination by the example coordination processor 1805. The example of fig. 19 illustrates travel data for three sites in a group of media sites (e.g., the same media site owner, the same media site type, etc.). The second column represents TAB DEC routing data for these sites and the third column represents routing data determined, for example, by the example routing processor 328 of fig. 3. The fourth column represents the traversal data after constraining the data to within +/-20% of the TABDEC traversal data. The fifth column represents the travel data after scaling to satisfy the constraint of keeping the average amount of travel constant. The sixth and seventh columns represent the resulting travel data after the second cycle. Finally, the last column represents the harmonized travel data that is relatively close in nature to the full convergence of the TAB DEC travel data while keeping the original overall average travel volume unchanged.
Returning to fig. 18, to estimate missing demographic data or information (e.g., responders may not specify ethnicity, language, job status, occupation, income, etc.), the statistics processing device 397 includes a data fusion processor 1810. Using any of a variety of data fusion techniques, the data fusion processor 1810 estimates or otherwise determines missing demographic data. The example data fusion processor 1810 of fig. 18 assumes that gender demographics are complete, accurate, and available to individual responders, and will use it as a linking variable or "hook" for data fusion. Other linking variables may be age, residence, etc., and are assumed to be available to the individual responders as well.
Data fusion techniques work on the principle that missing data of a responder can be estimated by a responder with similar characteristics. For example, two responders of similar age, gender, location, and occupation are statistically more likely to have similar income than randomly paired responders. Thus, missing data such as the revenue of a first responder can be reliably (statistically) estimated from the revenue of a second responder by finding responders with similar characteristics and sharing a common linked variable.
In an example, responders are divided into two groups within each category: recipients with at least one missing demographic data and donors with complete demographic records. Thus, four groups are created: #1 (male, donor), #2 (male, recipient), #3 (female, donor), and #4 (female, recipient). Next, statistical differences were calculated between each recipient and each donor across the various gender ranges. Finally, each recipient is paired with the donor with the least statistical difference, and demographic information from the donor is used for the recipient. It will be apparent to one of ordinary skill in the art that any suitable statistical difference may be employed. In the example, the statistical difference is calculated as the well-known improved Mahalanobis distance. This improved Mahalanobis distance is the distance between two N-dimensional points that are scaled according to the statistical differences, correlations, and importance of the various components of the points. In the example of FIG. 18, two N-dimensional points represent the recipient and donor's revenue per media site, where N is the number of media sites.
To generate a model suitable for characterizing the exposure or the impact (i.e., reach and frequency) of the focus (or media site type, owner, etc.) of a media site, the statistical processing means 397 includes a frequency and reach processor 1815. In an example, the frequency and reach processor 1815 determines parameters of a well-known Gamma Poisson distribution (i.e., a Negative Binomial Distribution (NBD)). In an example, the frequency and reach processor 1815 calculates a 9-day gross score (GRP) for a schedule (i.e., a set of media sites selected according to one or more criteria) based on the travel data. The frequency and coverage processor 1815 then estimates and obtains model parameters from the GRPs and coverage using any of a variety of well-known techniques. The frequency and reach processor 1815 then uses the estimated model parameters to rank the media effects (i.e., reach and frequency values) for a predetermined period of time. It will be apparent to those skilled in the art that other suitable models may be used; and other suitable models that can be used to determine model parameters and determine the reach and frequency values from the travel data can be used.
Fig. 20, 21, and 22 illustrate flow diagrams representative of example machine readable instructions that may be executed by a processor (e.g., one of the processors 2305A-C of fig. 23) to implement the data coordination processor 1805, the data fusion processor 1810, and the frequency and reach processor 1815, respectively. The machine readable instructions of the data coordination processor 1805, the data fusion processor 1810, and/or the data frequency and reach processor 1815 of fig. 20-22 may be executed by a processor, a controller, and/or a suitable processing device. For example, the machine readable instructions of the data coordination processor 1805, the data fusion processor 1810, and/or the frequency and reach processor 1815 of fig. 20-22 may be embodied as coded instructions stored on a tangible medium, such as flash memory, or Random Access Memory (RAM) associated with the processors 2305A-C shown in the example processor platform 2300 and discussed below in connection with fig. 23. In addition, some or all of the example machine readable instructions of fig. 20-22, the data coordination processor 1805, the data fusion processor 1810, and/or the frequency and reach processor 1815 may be implemented using an Application Specific Integrated Circuit (ASIC), a Programmable Logic Device (PLD), a Field Programmable Logic Device (FPLD), discrete logic, hardware, software, and/or firmware. Furthermore, some or all of the example machine readable instructions of fig. 20-22, the data coordination processor 1805, the data fusion processor 1810, and/or the frequency and reach processor 1815 may be implemented manually or as a combination of any of the foregoing techniques. Furthermore, although the example machine readable instructions of FIGS. 20-22 are described with reference to the flow of FIGS. 20-22, persons of ordinary skill in the art will readily appreciate that many other methods of implementing the data coordination processor 1805, the data fusion processor 1810, and/or the frequency and reach processor 1815, all of FIG. 18, may be employed. For example, the order of execution of the blocks may be changed, and/or some of the blocks described may be changed, eliminated, or combined.
The example machine readable instructions of FIG. 20 begin with the data coordination processor 1805 processing various media site groups (e.g., media site owner, media site type, etc.) (block 2005). For a media site group, the data coordination processor 1805 identifies all media sites belonging to the media site group (block 2010). The data coordination processor 1805 then merges the travel data by adding together the travel data of all respondents to each media site (block 2015). The data coordination processor 1805 then calculates an Average (AP) of the travel data for the identified media stations (block 2020). For each loop of data reconciliation (block 2030), the data reconciliation processor 1805 processes each media site in the media group (block 2035). For each media site (block 2035), the data coordination processor 1805 constrains the routing data corresponding to that site to within +/-20% of the TAB DEC routing data (block 2040). If there are more media sites in the group of media sites that have not been processed (block 2045), the data coordination processor 1805 returns to block 2035 to process the next media site in the group of media sites. Otherwise, if all of the media sites in the group of media sites have been processed (block 2045), the data coordination processor 1805 scales the travel data corresponding to the media sites in the group of media sites (as previously discussed) (block 2050).
Next, the data coordination processor 1805 determines (as discussed previously) whether the data coordination result for the traveling data has converged (block 2055). If convergence has occurred (block 2055), the data coordination processor 1805 proceeds to block 2065 to determine if all media site groups have been processed. Otherwise, if convergence has not occurred (block 2055), the data coordination processor 1805 returns to block 2030 to process the next loop.
If all of the media site groups have been processed (block 2055), the data coordination processor 1805 ends execution of the example machine readable instructions of FIG. 20. Otherwise, if there are more media site groups that have not been processed (block 2065), the data coordination processor 1805 returns to block 2005 to process the next media site group.
The example machine readable instructions of fig. 21 begin with the data fusion processor 1810 processing the respective gender (block 2105). Within each gender (block 2105), the data fusion processor 1810 classifies each responder 102 as either a donor or a recipient based on whether the responder has complete demographic information (block 2110). For each identified recipient (block 2115), the data fusion processor 1810 sets the current minimum value to a zero value (block 2117), and processes all donors of the same gender (block 2120). For one donor (block 2120), the data fusion processor 1810 calculates statistical differences between the recipient and the donor (as discussed previously) (block 2125). If the statistical difference is less than the current minimum value (block 2130), the data fusion processor 1810 notes the donor (i.e., records information identifying the donor) and updates the minimum value to be equal to the statistical difference (block 2135).
If the data fusion processor 1810 does not calculate the statistical difference for all donors of the same gender (block 2140), the data fusion processor 1810 returns to block 2120 to process the next donor. If all donors have been processed (block 2140), the data fusion processor 1810 populates the recipient's missing demographic information with information from the noted donors (block 2145). If the data fusion processor 1810 has not processed all recipients (block 2150), the data fusion processor 1810 returns to block 2115 to process the next recipient. If all recipients have been processed (block 2150) and none of the sexes have been processed (block 2155), the data fusion processor 1810 returns to block 2105 to process the next gender. If all genders have been processed (block 2155), the data fusion processor 1810 ends execution of the example machine readable instructions of FIG. 21.
The example machine readable instructions of fig. 22 begin with the range and frequency processor 1815 calculating a 9-day GRP for a predetermined nine-day group using coordinated traversal data corresponding to the predetermined media site group (i.e., schedule) (block 2205). The coverage and frequency processor 1815 then performs a weighted coverage and frequency analysis using the travel data corresponding to the responders reported on day 1 in the predetermined nine day group (block 2210). The output of this analysis is used to generate the initial parameters of the Gamma Poisson model. Next, the reach and frequency processor 1815 scales the initial model parameters to ensure that the model produces results that are consistent with the reconciled 9-day GRPs (block 2215), and uses the scaled parameters to calculate the desired reach and frequency data (block 2220).
Fig. 23 illustrates an example processor system 2300 capable of implementing the methods and apparatus disclosed herein. The processor system 2300 includes one or more processors 2305A-C with associated system memory. The system memory may include one or more of Random Access Memory (RAM)2315 and Read Only Memory (ROM) 2317.
In the example of FIG. 23, the multiple processors 2305A-C communicate with an input/output controller hub (ICH)2325 that interfaces with other peripherals or devices. In the example shown, peripherals that interface with the ICH 2325 include an input device 2327, a mass storage device 2340 (e.g., a hard disk drive), a Universal Serial Bus (USB)2345, a USB device 2350, a network port 2355 (which also communicates with a network 2360), and/or a removable storage device driver 2357. The removable storage device drive 2357 may include associated removable storage media 2358, such as magnetic media or optical media. One or more peripheral devices may enable the provision of the recorded positioning data 305 by the download server 120. The mass storage device 2340 may be used to store the example machine readable instructions illustrated in fig. 6A-6C, 15-17, and 20-22.
The example processor system 2300 of FIG. 23 also includes a video graphics adapter card 2320, which is a peripheral device that communicates with the Memory Controller Hub (MCH)2310 and also communicates with the display device 2322.
The example processor system 2300 may be, for example, a conventional desktop personal computer, a notebook computer, a workstation, a network server, or any other computing device. The processors 2305A-C may be any type of processing unit, such as Intel ® Pentium ® family of microprocessors, Intel ® Itanium ® family of microprocessors, Intel Xscale ® family of processors, AMD ® AthlonTMSerial processors and/or AMD ® OpteronTMA microprocessor of a series of processors. The processors 2305A-C may execute the example machine readable instructions shown in fig. 6A-6C, 15-17, and 20-22 to implement the MECD 300, the preprocessor 308, the travel path processor 310, the media site processor 320, the travel path processor 328, the statistics processor 397, the data coordination processor 1805, the data fusion processor 1810, and/or the frequency and reach processor 1815.
The memories 2315 and 2317 (which form part or all of the system memory) may be any suitable memory or storage device and may have capacities commensurate with the storage requirements of the system 2300. Furthermore, the mass storage device 2340 may be any magnetic or optical medium that is readable, for example, by the processors 2305A-C. The system memory may be used to store the recorded travel path data 305, the enhanced travel path data 315, the media site data 325, and/or the database 130. The system memory may also be used to store the example machine readable instructions shown in fig. 6A-6C, 15-17, and 20-22.
The input device 2327 may be implemented by a keyboard, mouse, touch screen, track pad, or any other device that enables a user to provide information to the processors 2305A-C.
The display device 2322 may be, for example, a Liquid Crystal Display (LCD) monitor, a Cathode Ray Tube (CRT) monitor, or any other suitable device that serves as an interface between the processors 2305A-C and a user via the video graphics adapter 2320. The video graphics adapter 2320 is any device used to connect the display device 2322 with the MCH 2310. These cards are currently commercially available from, for example, Creative Labs and other similar suppliers.
The removable storage device drive 2357 may be, for example, an optical drive such as a compact disk recordable (CD-R) drive, a compact disk rewritable (CD-RW) drive, a Digital Versatile Disk (DVD) drive, or any other optical drive. And may be, for example, a magnetic media drive. The removable storage media 2358 is a complementary product to the removable storage device drive 2357, and in this regard, the media 2358 is selected to work with the drive 2357. For example, if the removable storage device drive 2357 is an optical drive, the removable storage media 2358 may be a CD-R disk, a CD-RW disk, a DVD disk, or any other suitable optical disk. On the other hand, if the removable storage device drive 2357 is a magnetic media device drive, the removable storage media 2358 may be, for example, a diskette or any other suitable magnetic storage media. The removable storage media 2358 may also be used to provide recorded locations through the download server 120 or to store the database 130.
The example processor system 2300 also includes a network port 2355 (e.g., a processor peripheral), such as an ethernet card or any other card (which may be wired or wireless). The network port 2355 provides a network connection between the processors 2305A-C and a network 2360, which network 2360 may be a Local Area Network (LAN), Wide Area Network (WAN), the internet, or any other suitable network. The network port 2355 and network 2360 may also be used to provide recorded position fixes by the download server 120.
Of course, one of ordinary skill in the art will recognize that the order, size, and proportions of the memories shown in the example system may vary. Furthermore, although this patent discloses example systems including, among other components, software or firmware executed on hardware, it should be noted that such systems are merely illustrative and should not be considered as limiting. For example, it is contemplated that any or all of these hardware and software components could be embodied exclusively in hardware, exclusively in software, exclusively in firmware or in some combination of hardware, firmware and/or software. Thus, those of ordinary skill in the art will readily recognize that the above-described examples are not the only way to implement such systems.
At least some of the above-described example methods, machine readable instructions, and/or apparatus are implemented by one or more software and/or firmware programs running on a computer processor. However, dedicated hardware implementations including, but not limited to, application specific integrated circuits, programmable logic arrays and other hardware devices can likewise be constructed to implement some or all of the example methods and/or apparatus described herein, in whole or in part. Moreover, other alternative software implementations including, but not limited to, distributed processing or component/object distributed processing, parallel processing, or virtual machine processing can also be constructed to implement the example methods and/or apparatus described herein.
It should also be noted that the example software and/or firmware implementations described herein are stored as needed in tangible storage media, such as: magnetic media (e.g., magnetic disks or tapes); magneto-optical or optical media such as a disc; or a solid-state medium such as a memory card or other form of packaging that houses one or more read-only (non-volatile) memories, random access memories, or other re-writable (volatile) memories; or a signal containing computer instructions. Digital file attachments to e-mail or other self-contained information documents or document sets are considered a distribution medium equivalent to a tangible storage medium. Thus, the example software and/or firmware described herein may be stored on a tangible storage medium or distribution medium such as those described above or an equivalent or later developed medium.
From the standpoint of the foregoing description being exemplary components and functions described with reference to certain standards and protocols, it should be understood that the teachings of this disclosure are not limited to such standards and protocols. For example, various standards for internet and other packet switched network transmissions (e.g., Transmission Control Protocol (TCP)/IP, User Datagram Protocol (UDP)/IP, hypertext markup language (HTML), hypertext transfer communication protocol (THHP)), and various standards for inter-computer and inter-device communications (e.g., USB) are examples of the prior art. These standards are being periodically replaced by faster and more efficient equivalent protocols with the same overall functionality. Accordingly, replacement standards and protocols having equivalent functions that are anticipated by the teachings of this disclosure are intended to be included within the scope of the following claims.
The teachings of the present disclosure contemplate one or more machine-readable media containing instructions or receiving and executing instructions from a propagated signal so that, for example, a device connected to a network environment can use the instructions to send or receive voice, video, or data over a network and communicate. Such devices may be implemented by any electronic device that provides voice, video, or data communication, such as a telephone, cordless telephone, mobile telephone, cellular telephone, Personal Digital Assistant (PDA), set-top box, computer, and/or server.
Although certain example methods, apparatus and articles of manufacture have been described herein, the scope of coverage of this patent is not limited thereto. On the contrary, this patent covers all methods, apparatus and articles of manufacture fairly falling within the scope of the appended claims either literally or under the doctrine of equivalents.
The present invention claims the benefits of U.S. provisional application serial No. 60/592554 entitled Methods and Apparatus for Processing Data Collected by a GPS-enabled Media Measurement system filed on 30.7.2004, U.S. provisional application serial No. 60/681785 filed on 17.5.2005, U.S. provisional application serial No. 60/681785 filed on the 17.5.2005, and U.S. provisional application serial No. 60/688494 filed on 8.6.2005, entitled Methods and Apparatus for Improving the access and Reach of Electronic Media Exposure Measurement Systems. U.S. provisional applications serial nos. 60/592554, 60/681785, 60/688494 and U.S. applications serial nos. 10/686872 and 10/318422 are hereby incorporated by reference in their entirety.