A kind of market shopping path recommended method and deviceTechnical field
The present invention relates to data processing field more particularly to a kind of market shopping path recommended method and devices.
Background technique
In megastore, shop is very more, and each layer of length is up to several hundred rice.And shop title is varied very muchEnglish name, people are not likely to be familiar with the title in each shop and business of doing business.So if there is personalized market to stroll shopIt guides, for the stronger customer of Objective that does shopping, it can save customer and buy the time of article, and can save and repeat walkingEnergy, in addition can also avoid omitting valuable shop.The present invention, can be to customer by wifi location fingerprint technologyProperty shopping guide.
The present invention is based on wifi location fingerprint is based on, the shop categorical data that customer enters is collected, is closed according to shop typeConnection carries out collaborative filtering.It is successively sorted according to market layout and positional relationship, according to customer's floor, recommends customer from the near to the distantRelated shop has been strolled, and relevant advertisements can be beaten.
Summary of the invention
The present invention provides a kind of market shopping path recommended method and devices, for helping customer recommendation in emporiumTo the helpful shop of customer, customer is avoided to can not find shop, or repeats to take a long walk.
The present invention provides a kind of market shopping path recommended methods, mainly comprise the steps that
1, a kind of market shopping path recommended method, which is characterized in that the described method includes:
The address customer's mobile phone mac is obtained by market wifi；
The position in the market shop Zhong Ge is determined according to wifi location fingerprint；
According to wifi location fingerprint, determines the store location of user's access, location error is handled；
According to the record in the shop that customer visited, the interested shop of user is recommended, obtains and recommends shop；
The current distance for recommending shop and customer is calculated, optimization customer reaches the path in shop；
Customer is guided to reach the recommendation shop.
2, according to the method described in claim 1, wherein, the address acquisition customer mobile phone mac specifically includes that
To the mobile phone for being linked into market wifi, its body is determined by reading the address MAC in its communication process in data framePart；
It, can be when it carries out authentication if mobile phone uses drive sweep mode to the mobile phone for not being linked into market wifiThe address MAC in Authentication frame is read, determines its identity；
It, can be by reading its Probe if mobile phone uses active scan mode to the mobile phone for not being linked into market wifiThe address MAC in request frame, determines its identity.
3, described according to wifi location fingerprint according to the method described in claim 1, wherein, determine the market shop Zhong GePosition, specifically include that
Acquisition signal strength establishes signal strength fingerprint base, by matching real-time signal strength collected at point to be determinedReference point locations fingerprint in fingerprint base obtains wifi location fingerprint；
It is positioned by the location fingerprint of wifi, determines the position in shop.
4, described that location error is handled according to the method described in claim 1, wherein, it specifically includes that
Since there are wifi position errors, when the position that customer is positioned is between two shops, selecting customer more to have canThe shop that can be gone；The target shop that customer subsequently accesses is inferred in the shop gone according to customer；Before access shopSimilitude between afterwards infers the access path of customer.
5, according to the method described in claim 1, wherein, the record in the shop visited according to customer feels userThe shop of interest is recommended, and is obtained and is recommended shop, specifically includes that
When customer enters market, and after the residence time reaches certain threshold value in shop, the shop and relevant bits confidence are recordedBreath；The n shop that record target customers enters in market；
According to all customers to the preference of article or information, finds customer base similar with target customers's demand and preference, adoptCustomer's cluster is carried out with " K- arest neighbors " algorithm；Then, the shop based on this K most similar customers accesses preference information,Shop recommendation is carried out for target customers.
6, according to the method described in claim 1, wherein, the current distance for calculating recommendation shop and customer optimizesCustomer reaches the path in shop, specifically includes that
After obtaining the store information recommended, according to wifi location fingerprint information, position of these shops in market is calculated, and tieFloor where closing customer is searched for closest to the shop of customer in destination, and recommends to reach target shop from current locationWalking path gives target customers.
Still optionally further, in method as described above, the guide customer reaches shop, specifically includes that
Direct customers are walked to nearest recommendation shop, and shop relative information displaying to customer, customer can choose whether to connectBy associated recommendation, if not liking the recommendation, next recommendation results can also be further selected.
The present invention provides a kind of market shopping path recommendation apparatus, described device includes:
Mobile phone mac address acquisition module, for obtaining the address customer's mobile phone mac by wifi；
Store location locating module, for being positioned according to position of the wifi location fingerprint to shop；
Customer location error concealment module, judges for the position to customer, avoids the positioning of mistake；
Collaborative filtering module, the shop for being strolled according to customer, recommends relevant shop；
Shop recommending module optimizes the path in decorrelation shop for the position according to customer, and carries out pushing away for arcade shop premisesIt recommends.
Technical solution provided in an embodiment of the present invention can include the following benefits:
The present invention can recommend related shop, and successively sort according to market layout and positional relationship, root according to the demand of customerAccording to customer's floor, customer is recommended to stroll related shop from the near to the distant.For the stronger customer of Objective that does shopping, it can saveCustomer buys the time of article, and can save the energy for repeating walking, in addition can also avoid omitting valuable shop.
Detailed description of the invention
Fig. 1 is the flow chart of market shopping path recommended method embodiment of the invention；
Fig. 2 is the structure chart of market shopping path recommendation apparatus embodiment of the invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, right in the following with reference to the drawings and specific embodimentsThe present invention is described in detail.
Fig. 1 is the flow chart of market shopping path recommended method embodiment of the invention.As shown in Figure 1, the present embodiment oneKind market shopping path recommended method, can specifically include following steps:
Step 101, active wifi information source monitoring, records the mobile phone mac of customer.
Universal and market wifi with smart mobile phone application is popularized.More and more markets both provide free nothingLine WiFi service, customer's smart phone is essentially all a work station.Due to each work station have one it is globally unique48 address medium access control (Medium Access Control, MAC), and each frame issued includesThere is the address MAC, therefore the AP in network can carry out WiFi to work station by reading the address MAC in frameIdentification.The method of identification can be divided into following three kinds according to different situations:
It 1), can be by reading the address MAC in its communication process in data frame come really to the work station being linked into Home NetworkIts fixed identity.
2) to the work station not being linked into Home Network, if work station uses drive sweep mode, body can be carried out at itThe address MAC in Authentication frame is read when part verifying, determines its identity.In this way, even if the workMake station without the permission of access present networks, its identity can also be identified.
It 3), can be by reading it if work station uses active scan mode to the work station not being linked into Home NetworkThe address MAC in Probe request frame, determines its identity.In this way, even if the work station does not access Home NetworkNetwork can also identify its identity.
In practical work process, identification can be carried out to work station using above-mentioned three kinds of methods simultaneously, to determine workWhether in a network to stand.Its workflow is as follows: firstly, need to establish a MAC address list in AP, then APUtilize the address MAC in the continuous read data frame of above-mentioned three kinds of methods.If reading the address MAC not in lists,Think there is new work station to enter network, needs for be added in list the address MAC at this time;If MAC has been deposited addressBe in list, then it is believed that corresponding work station in a network;If some address MAC in list is in a period of timeIt is never read into (such as 20 seconds), then it is believed that corresponding work station has left network, is needed at this time by thisThe address MAC is removed from the list.
Step 102, design position fingerprint is aligned the store location in market with the position wifi.
WIFI technology is based on IEEE802.11 agreement, works in 2.4GHz/5GHz. WIFI with signal and covers modelEnclose the advantages that wide, transmission speed is fast, customer facilitates access.It is mainly utilized based on WIFI indoor positioning technologies without space bitSet has highly relevant property between signal receiving strength, includes two implementation phases based on location fingerprint library method.First fromLine phase acquisition signal strength establishes signal strength fingerprint base, On-line matching stage, by real-time letter collected at point to be determinedNumber intensity matches the reference point locations fingerprint in fingerprint base by learning algorithm, and this method position error is on 2 meters to a 4 meters left sidesIt is right.This localization method advantage is to require no knowledge about accurate signal attenuation model.
The Similarity measures and store location Error processing of step 103 customer location error.
The position range error in shop is also possible that error at 2 meters to 4 meters.Therefore the present invention, exists also according to customerThe shop that front was gone.It is inferred to, what shop the shop of customer's access is.By the similitude between shop, target is inferredThe access path of customer.Such as when customer has accessed a clothes shop.The positioning target position data that system subsequently records is aobviousShow, he between clothes shop and pet food shop, then system according to front with the similitude in clothes shop, clothes shop can be selected,The record in shop is accessed as customer.And when the position that be identified based on wifi location fingerprint, be not between two shops, orDiffer less than half meter；And be biased into, it is then that customer accesses shop with the position in the deviation shop when some shop.System is notThe disconnected store information accessed these users records.
Step 103, position positioning is carried out to the customer in market, and according to wifi location fingerprint information, they is stored inIn database.
When new customer enters market, into n shop, and after the residence time reaches certain threshold value in shop, recordThe shop and relevant position.This process can be in recommender system with classification, and customer is added to article.Pass through this typeThan, it will be able to information recommendation is carried out based on the method for recommender system.
Step 104, the position in customer and shop carries out the calculating of floor and distance, according to several before customer entranceShop record, carries out collaborative filtering.
The basic principle of collaborative filtering recommending based on customer is, according to all customers to the preference of article or information,It was found that " neighbours " customer base similar with target customers's taste and preference, is the algorithm for using " K- neighbours " in general applicationTo calculate the phase recency between customer；Then, based on the history preference information of this K neighbour, information is carried out for target customersRecommend.
It, can he strolled not yet to customer recommendation shop by collaborative filtering.And according to former customer'sBehavioural information calculates in the shop strolled with oneself, and the most similar customer of behavior predicts that this customer is also possible to that shop can be strolled, i.e.,Calculating similar customers can be more long in what shop stop, and system recommends relevant shop to it.
Step 105, distance calculates, and optimization goes to the path in the shop.Obtain the information in the shop that these may go to strollAfterwards, according to wifi location fingerprint information, positional distance of these shops in market is calculated, and combines floor where customer, ifIn the destination that meter needs to recommend, closest to the shop of customer, the path that target shop is reached from target position is designed.
Step 106, when the equipment such as customer's mobile phone or ipad carry out when connecting market wifi or using correlation appShop is recommended and route guiding, direct customers are walked to nearest recommendation shop, and shop relevant information, for example, advertisement, it is preferential,The information such as discount show customer by equipment such as mobile phone or ipad.Customer can choose whether to receive associated recommendation, ifThe recommendation is not liked, next recommendation results can also be further selected.The further direct customers of system are walked to new shop.It savesThe time that customer finds back and forth, and can the important information in shop be more presented to customer.It avoids important and usefulIt is missed by customer in shop.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment canThe mode of necessary general hardware platform can also be added to realize by software by software realization.Based on this understanding,The technical solution of above-described embodiment can be embodied in the form of software products, which can store non-easy at oneIn the property lost storage medium (can be CD-ROM, USB flash disk, mobile hard disk etc.), including some instructions are with so that a computer is setStandby (can be personal computer, server or the network equipment etc.) executes method described in each embodiment of the present invention.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the inventionWithin mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.