US20170011025A1 - Sentence simplification for spoken language understanding - Google Patents

Abstract

Sentence simplification may be provided. A spoken phrase may be received and converted to a text phrase. An intent associated with the text phrase may be identified. The text phrase may then be reformatted according to the identified intent and a task may be performed according to the reformatted text phrase.

Description

CROSS REFERENCE TO RELATED APPLICATION
• The present application is a continuation of U.S. patent application Ser. No. 13/106,374, filed May 12, 2011, which application is hereby incorporated by reference in its entirety.
BACKGROUND
• Sentence simplification is a process for improving the robustness of spoken language understanding tasks. In conventional systems, simple well-formed sentences may be converted accurately, but the error rate increases as the sentences become more complex. For example, longer, natural, and/or spontaneous utterances often prove difficult to convert and result in inaccurate processing. In some situations, the speaker's intent may be reversed by a negation. Conventional systems depend on word n-grams where n is less than 5; that is, such systems cannot discern intent changes when the changing word is more than five words away in the original sentence.
SUMMARY
• This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter. Nor is this Summary intended to be used to limit the claimed subject matter's scope.
• Sentence simplification may be provided. A spoken phrase may be received and converted to a text phrase. An intent associated with the text phrase may be identified. The text phrase may then be reformatted according to the identified intent and a task may be performed according to the reformatted text phrase.
• Both the foregoing general description and the following detailed description provide examples and are explanatory only. Accordingly, the foregoing general description and the following detailed description should not be considered to be restrictive. Further, features or variations may be provided in addition to those set forth herein. For example, embodiments may be directed to various feature combinations and sub-combinations described in the detailed description.
BRIEF DESCRIPTION OF THE DRAWINGS
• The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate various embodiments of the present invention. In the drawings:
• FIG. 1 is a block diagram of an operating environment;
• FIG. 2 is an illustration of a dependency parse of an example sentence;
• FIG. 3 is a flow chart of a method for providing sentence simplification; and
• FIG. 4 is a block diagram of a system including a computing device.
DETAILED DESCRIPTION
• The following detailed description refers to the accompanying drawings. Wherever possible, the same reference numbers are used in the drawings and the following description to refer to the same or similar elements. While embodiments of the invention may be described, modifications, adaptations, and other implementations are possible. For example, substitutions, additions, or modifications may be made to the elements illustrated in the drawings, and the methods described herein may be modified by substituting, reordering, or adding stages to the disclosed methods. Accordingly, the following detailed description does not limit the invention. Instead, the proper scope of the invention is defined by the appended claims.
• Sentence simplification for spoken language understanding (SLU) may be provided. Consistent with embodiments of the present invention, a dependency parsing-based sentence simplification approach may extract a set of keywords from natural language sentences. Those keywords may be used in association with the complete sentences in order to improve the accuracy of SLU tasks.
• Speech-to-text conversion (i.e., speech recognition) may comprise converting a spoken phrase into a text phrase that may be processed by a computing system. Acoustic modeling and/or language modeling may be used in modern statistic-based speech recognition algorithms. Hidden Markov models (HMMs) are widely used in many conventional systems. HMMs may comprise statistical models that may output a sequence of symbols or quantities. HMMs may be used in speech recognition because a speech signal may be viewed as a piecewise stationary signal or a short-time stationary signal. In a short-time (e.g., 10 milliseconds), speech may be approximated as a stationary process.
• Syntactic information may be used to preprocess spoken language inputs and prepare the feature set for SLU tasks. This may assist in alleviating the problem that paths in a parse tree may occur a relatively small number of times. A simple negation, for example, may totally change the structure of the syntactic parse tree. Sentence simplification may then help this problem by condensing the training and test sets so that the classifier will work better as the average frequency of candidate lexical and syntactic features increase.
• Further, long distance dependencies may be better handled without bombarding the classifier with candidate syntactic features. Sentence simplification may thus comprise an utterance compression task with a goal of rephrasing the same intent with fewer words and supporting short, keyword sequence inputs. This may be analogous to understanding keyword-based queries where there is usually a natural language query in mind. For example the query “What is the capacity of a 737” may be rephrased as “capacity 737”.
• While sentence simplification makes sense for intent determination, which is typically framed as an utterance classification task, this approach may also be effective for slot filling due to its power for handling long distance dependencies. For both tasks, the approach may rely on features extracted from the dependency parse of the input utterance.
• FIG. 1 is a block diagram of anoperating environment100.Operating environment100 may comprise, for example, auser device110 such as a computer, laptop, tablet, cellular phone, and/or other electronic device.User device110 may comprise an input device120 (e.g., a keyboard, mouse, and/or touch screen), akeyword database125 stored in a memory, amicrophone130, and adisplay140.
• FIG. 2 is an illustration of a dependency parse of anexample sentence200.Example sentence200 may comprise a plurality of words received viamicrophone130, in this case “I need to fly from Boston to London.” Each word may comprise a dependency on at least one other word and/or may be associated with at least one other word that depends on it. For example, “I” comprises afirst dependency210 on “need”, “need” comprises asecond dependency220 on “to”, and “to” comprises athird dependency230 on “fly”. The second “to” may comprise afourth dependency240 on “fly” and “from” may comprise afifth dependency250 on “fly”. “Boston” may comprise asixth dependency260 on “from” and “London” may comprise aseventh dependency270 on the second “to”.
• In the example ofFIG. 2, “fly” may comprise a top level predicate. The top level predicate may comprise a word on which others depend, but which does not depend on any other words itself. Words that depend on the top level predicate may comprise auxiliary words that may be ancillary to the speaker's intent. For example “I” and “need” may be excluded from the intent determination as auxiliary words as the intended task may be implicitly understood to be associated with something the speaker needs or wants to accomplish. Such auxiliary words may be predefined, such as inkeyword database125. Prepositions such as “from” and/or “to” may be retained as they may be salient to identifying the user's intent. For example, “I need to fly from Boston to London” may comprise an identified intent of booking a flight based on a reformatted instruction phrase of “fly from to” while “How much does it cost to fly from Boston to London” may comprise an identified intent of looking up a fare based on a reformatted instruction phrase of “cost to fly.”
• FIG. 3 is a flow chart setting forth the general stages involved in amethod300 consistent with an embodiment of the invention for providing sentence simplification.Method300 may be implemented using acomputing device400 as described in more detail below with respect toFIG. 4. Ways to implement the stages ofmethod300 will be described in greater detail below.Method300 may begin at startingblock305 and proceed to stage310 wherecomputing device400 may perform a dependency parse on a text phrase. For example, the text phrase “I need to fly from Boston to London” may be parsed such that each word of the phrase depends on and/or is depended upon by at least one other word of the phrase. Consistent with embodiments of the invention, the text phrase may comprise a spoken phrase received from a user and converted to text.
• The structure of a sentence may determined by the relation between a word (a head) and its dependents. Each word may point to a head; for example, for the noun phrase “blue book”, blue points to book. A parser may be trained from a treebank following a latent variable approach by iteratively splitting non-terminals to better represent the data. A Constituency-to-Dependency Conversion toolkit may be used to form dependency parses from the output parse trees. To adapt the parser to the speech domain, a self-training approach using a training data set, such as the Airline Travel Information System (ATIS) project data set, may be used. The ATIS data set is described in “Evaluation of spoken language systems: The ATIS domain,” by P. J. Price as published in Proceedings of the DARPA Workshop on Speech and Natural Language, Hidden Valley, Pa., June 1990, which is hereby incorporated by reference in its entirety.
• Fromstage310,method300 may advance to stage315 wherecomputing device400 may identify a top level predicate. The top level predicate may comprise a word of the text phrase upon which other words depend but which does not depend on any other words itself. For example, in the phrase “I need to fly from Boston to London”, “fly” may comprise the top level predicate. The top level predicate may thus comprise the critical word to the user's intent.
• Fromstage315,method300 may advance to stage320 wherecomputing device400 may reformat the text phrase. For example, the natural language text phrase “I need to fly from Boston to London” may be reformatted as “fly to”. The text phrase “I want to make a dinner reservation at Mario's for 6 people at 7 o'clock” may be reformatted as “reservation at”.
• Icsiboost2, an implementation of the AdaBoost.MH algorithm, may be used for intent identification. Boosting is an iterative procedure that builds a new weak learner h_tat each iteration. Each example of the training data set may be assigned a weight. These weights may initialized uniformly and updated on each iteration so that the algorithm focuses on the examples that were wrongly classified on the previous iteration. At the end of the learning process, the weak learners used on each iteration t may be linearly combined to form the classification function shown inEquation 1. α_tmay comprise a weight of the weak learner h_tand T comprises the number of iterations of the algorithm.
• $\begin{array}{cc}f\left(x,l\right)=\sum _{t=1}^T{\alpha }_th_t\left(x,l\right)& \mathrm{Equation}1\end{array}$
• Fromstage320,method300 may advance to stage325 wherecomputing device400 may define an intent domain according to the reformatted phrase. For example, a reformatted text phrase of “fly to” may be associated with a travel domain while “reservation at” may be associated with a dining domain.
• Fromstage325,method300 may advance to stage330 wherecomputing device400 may fill at least one semantic slot. Each intent domain may be associated with a plurality of semantic slots. For example, a travel domain may comprise slots for departure location, departure time, arrival time, arrival destination, travel method, constraints (e.g., lowest cost, no layovers, etc.). A dining domain may comprise semantic slots for cuisine type, time, party size, reservation needed, relative location (e.g., in walking distance, near a transit station, etc.). The semantic slots may be filled with words from the text phrase and/or implied words. Slots may be filled with words that may have been excluded from the reformatted phrase. For example, the phrase “I need to fly from Boston to London” may result in filling a semantic slot for departure location with “Boston” and a semantic slot for arrival location with “London”.
• Slot filling may need to handle long distance dependencies between the slot word or phrase and its disambiguator. A disambiguator may comprise a phrase that determines a semantic subcategory of an entity. For example, the word “morning” may be known to be a time period. The semantic disambiguation of whether it is an arrival or departure time within a travel domain intent may rely on a dependent predicate, such as arriving. Similar to intent determination, the slot filler may exploit the phrase dependency structures while determining the semantic slot. It is easy to detect a day name keyword, such as “Saturday,” but to understand whether it is departure, arrival, or return day may rely on not only the top level predicate but predicates of other clauses. For the example sentence “Find flights departing from New York tomorrow arriving in Tokyo no later than Saturday,” the predicate “arrive” may be considered as a feature while classifying the words which directly or indirectly depend on it (e.g., “Tokyo” and “no later than Saturday”). The recursive algorithm to find the predicate head of a given word may be as follows: if the head of a word is a predicate, then it is used, otherwise, the predicate head of its head is used as its predicate head.
• Fromstage330,method300 may advance to stage335 wherecomputing device400 may perform a task. For example, the defined domain and semantic slots may be used to perform an Internet search, make dining reservations, purchase airline travel tickets, and/or create a meeting request.
• Fromstage335,method300 may advance to stage340 wherecomputing device400 may display a result to the user. For example, an original phrase of “I need to fly from Boston to London” may result in searching for airline tickets and displaying a plurality of options such as different airlines, times, and/or prices.Method300 may then end atstage345.
• An embodiment consistent with the invention may comprise a system for providing sentence simplification. The system may comprise a memory storage and a processing unit coupled to the memory storage. The processing unit may be operative to receive a spoken utterance and/or phrase, convert the spoken phrase to a text phrase, identify an intent—which may be defined by a top level predicate—associated with the text phrase, reformat the text phrase according to the identified intent, and perform a task according to the reformatted text phrase. Being operative to identify the intent associated with the text phrase may comprise being operative to perform a dependency parse on the text phrase such as by identifying a top level predicate and excluding at least one auxiliary word, which may comprise a dependent of the top level predicate, in the text phrase. For example, in the dependency parse ofexample sentence200, “Boston” and “London” are auxiliary dependents of “from” and “to”, respectively, which are in turn dependents of the top level predicate “fly”. “Boston”, “London”, and “from” may be excluded from the reformatted phrase “fly to”. The auxiliary keyword may also comprise a predefined auxiliary keyword such as “need” and/or “want” that may be automatically excluded. Identifying the top level predicate comprises evaluating a statistical weighting criterion associated with each word of the text phrase. For example, “What is the cost to fly from Boston to London” may result in a top level predicate of “cost” while “I need to fly from Boston to London” results in a top level predicate of “fly”. Although both phrases comprise the word “fly”, in the former example “cost” may comprise a higher statistical weighting indicating that it is more likely to define the user's intent. Reformatting the text phrase according to the identified intent may comprise defining a domain associated with the task and/or filling at least one semantic slot associated with the defined domain.
• Another embodiment consistent with the invention may comprise a system for providing sentence simplification. The system may comprise a memory storage and a processing unit coupled to the memory storage. The processing unit may be operative to identify an intent associated with a text phrase, wherein the text phrase comprises a plurality of words, reformat the text phrase according to the identified intent, perform a task according to the reformatted text phrase, and display a result associated with the task to a user. The text phrase may comprise a conversion of a spoken phrase received from the user and may be associated with a search query. Identifying the intent may comprise defining a target domain and the reformatted text phrase may comprise at least one keyword associated with the target domain. The target domain may comprises a plurality of semantic slots and the processing unit may be operative to fill at least one of the semantic slots with at least one slot word of the plurality of words. Filling the at least one of the semantic slots may comprise associating the at least one slot word of the plurality of words with the at least one of the semantic slots according to at least one second word of the plurality of words, wherein the at least one second word may comprise a dependent word associated with the at least one slot word during the dependency parse.
• Yet another embodiment consistent with the invention may comprise a system for providing sentence simplification. The system may comprise a memory storage and a processing unit coupled to the memory storage. The processing unit may be operative to receive a spoken phrase from a user, convert the spoken phrase to a text phrase, wherein the text phrase comprises a plurality of words, perform a dependency parse on the text phrase, identify a top level predicate of the text phrase, reformat the text phrase according to the top level predicate, wherein the reformatted text phrase excludes at least one first word of the plurality of words, define an intent domain according to the reformatted text phrase, wherein the defined intent comprises a plurality of semantic slots, fill at least one of the semantic slots with the at least one first word, wherein the at least one first word is associated with the at least one semantic slot according to at least one dependent word of the at least one first word, perform a task according to the defined intent and the filled at least one semantic slot, and display a result of the performed task to the user.
• FIG. 4 is a block diagram of a system includingcomputing device400. Consistent with an embodiment of the invention, the aforementioned memory storage and processing unit may be implemented in a computing device, such ascomputing device400 ofFIG. 4. Any suitable combination of hardware, software, or firmware may be used to implement the memory storage and processing unit. For example, the memory storage and processing unit may be implemented withcomputing device400 or any ofother computing devices418, in combination withcomputing device400. The aforementioned system, device, and processors are examples and other systems, devices, and processors may comprise the aforementioned memory storage and processing unit, consistent with embodiments of the invention. Furthermore,computing device400 may compriseoperating environment100 as described above.System100 may operate in other environments and is not limited tocomputing device400.
• With reference toFIG. 4, a system consistent with an embodiment of the invention may include a computing device, such ascomputing device400. In a basic configuration,computing device400 may include at least oneprocessing unit402 and asystem memory404. Depending on the configuration and type of computing device,system memory404 may comprise, but is not limited to, volatile (e.g., random access memory (RAM)), non-volatile (e.g., read-only memory (ROM)), flash memory, or any combination.System memory404 may includeoperating system405, one ormore programming modules406, and may includekeyword database125.Operating system405, for example, may be suitable for controllingcomputing device400's operation. In one embodiment,programming modules406 may include asentence parsing application407. Furthermore, embodiments of the invention may be practiced in conjunction with a graphics library, other operating systems, or any other application program and is not limited to any particular application or system. This basic configuration is illustrated inFIG. 4 by those components within a dashed line408.
• Computing device400 may have additional features or functionality. For example,computing device400 may also include additional data storage devices (removable and/or non-removable) such as, for example, magnetic disks, optical disks, or tape. Such additional storage is illustrated inFIG. 4 by a removable storage409 and anon-removable storage410.Computing device400 may also contain a communication connection416 that may allowdevice400 to communicate withother computing devices418, such as over a network in a distributed computing environment, for example, an intranet or the Internet. Communication connection416 is one example of communication media.
• The term computer readable media as used herein may include computer storage media. Computer storage media may include volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information, such as computer readable instructions, data structures, program modules, or other data.System memory404, removable storage409, andnon-removable storage410 are all computer storage media examples (i.e., memory storage.) Computer storage media may include, but is not limited to, RAM, ROM, electrically erasable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store information and which can be accessed by computingdevice400. Any such computer storage media may be part ofdevice400.Computing device400 may also have input device(s)412 such as a keyboard, a mouse, a pen, a sound input device, a touch input device, etc. Output device(s)414 such as a display, speakers, a printer, etc. may also be included. The aforementioned devices are examples and others may be used.
• The term computer readable media as used herein may also include communication media. Communication media may be embodied by computer readable instructions, data structures, program modules, or other data in a modulated data signal, such as a carrier wave or other transport mechanism, and includes any information delivery media. The term “modulated data signal” may describe a signal that has one or more characteristics set or changed in such a manner as to encode information in the signal. By way of example, and not limitation, communication media may include wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, radio frequency (RF), infrared, and other wireless media.
• As stated above, a number of program modules and data files may be stored insystem memory404, includingoperating system405. While executing onprocessing unit402, programming modules406 (e.g., sentence parsing application407) may perform processes including, for example, one or more ofmethod300's stages as described above. The aforementioned process is an example, andprocessing unit402 may perform other processes. Other programming modules that may be used in accordance with embodiments of the present invention may include electronic mail and contacts applications, word processing applications, spreadsheet applications, database applications, slide presentation applications, drawing or computer-aided application programs, etc.
• Generally, consistent with embodiments of the invention, program modules may include routines, programs, components, data structures, and other types of structures that may perform particular tasks or that may implement particular abstract data types. Moreover, embodiments of the invention may be practiced with other computer system configurations, including hand-held devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, minicomputers, mainframe computers, and the like. Embodiments of the invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.
• Furthermore, embodiments of the invention may be practiced in an electrical circuit comprising discrete electronic elements, packaged or integrated electronic chips containing logic gates, a circuit utilizing a microprocessor, or on a single chip containing electronic elements or microprocessors. Embodiments of the invention may also be practiced using other technologies capable of performing logical operations such as, for example, AND, OR, and NOT, including but not limited to, mechanical, optical, fluidic, and quantum technologies. In addition, embodiments of the invention may be practiced within a general purpose computer or in any other circuits or systems.
• Embodiments of the invention, for example, may be implemented as a computer process (method), a computing system, or as an article of manufacture, such as a computer program product or computer readable media. The computer program product may be a computer storage media readable by a computer system and encoding a computer program of instructions for executing a computer process. The computer program product may also be a propagated signal on a carrier readable by a computing system and encoding a computer program of instructions for executing a computer process. Accordingly, the present invention may be embodied in hardware and/or in software (including firmware, resident software, micro-code, etc.). In other words, embodiments of the present invention may take the form of a computer program product on a computer-usable or computer-readable storage medium having computer-usable or computer-readable program code embodied in the medium for use by or in connection with an instruction execution system. A computer-usable or computer-readable medium may be any medium that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
• The computer-usable or computer-readable medium may be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium. More specific computer-readable medium examples (a non-exhaustive list), the computer-readable medium may include the following: an electrical connection having one or more wires, a portable computer diskette, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CD-ROM). Note that the computer-usable or computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
• Embodiments of the present invention, for example, are described above with reference to block diagrams and/or operational illustrations of methods, systems, and computer program products according to embodiments of the invention. The functions/acts noted in the blocks may occur out of the order as shown in any flowchart. For example, two blocks shown in succession may in fact be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality/acts involved.
• While certain embodiments of the invention have been described, other embodiments may exist. Furthermore, although embodiments of the present invention have been described as being associated with data stored in memory and other storage mediums, data can also be stored on or read from other types of computer-readable media, such as secondary storage devices, like hard disks, floppy disks, or a CD-ROM, a carrier wave from the Internet, or other forms of RAM or ROM. Further, the disclosed methods' stages may be modified in any manner, including by reordering stages and/or inserting or deleting stages, without departing from the invention.
• All rights including copyrights in the code included herein are vested in and the property of the Applicant. The Applicant retains and reserves all rights in the code included herein, and grants permission to reproduce the material only in connection with reproduction of the granted patent and for no other purpose.
• While the specification includes examples, the invention's scope is indicated by the following claims. Furthermore, while the specification has been described in language specific to structural features and/or methodological acts, the claims are not limited to the features or acts described above. Rather, the specific features and acts described above are disclosed as example for embodiments of the invention.

Claims (20)

What is claimed is:

1. A method for providing sentence simplification, the method comprising:

receiving a spoken utterance;

converting the spoken utterance to a text phrase;

identifying a top level predicate associated with the text phrase;

reformatting the text phrase according to the identified predicate; and

performing a task according to the reformatted text phrase.

2. The method ofclaim 1, wherein identifying the top level predicate associated with the text phrase comprises performing a dependency parse on the text phrase.

3. The method ofclaim 2, wherein performing a dependency parse comprises:

identifying a top level predicate; and

excluding at least one auxiliary word in the text phrase.

4. The method ofclaim 3, wherein the at least one auxiliary word comprises a dependent of the top level predicate.

5. The method ofclaim 3, wherein the at least one auxiliary word comprises at least one predefined auxiliary keyword.

6. The method ofclaim 3, wherein identifying the top level predicate comprises evaluating a weighting criterion associated with each word of the text phrase.

7. The method ofclaim 1, wherein reformatting the text phrase according to the identified predicate comprises defining a domain associated with the task.

8. The method ofclaim 7, further comprising filling at least one semantic slot associated with the defined domain.

9. The method ofclaim 8, wherein the slot is filled with at least one word of the text phrase.

10. The method ofclaim 9, wherein the at least one word of the text phrase is not associated with the reformatted text phrase.

11. A computer-readable medium which stores a set of instructions which when executed performs a method for providing sentence simplification, the method executed by the set of instructions comprising:

identifying a predicate associated with a text phrase, wherein the text phrase comprises a plurality of words;

reformatting the text phrase according to the identified predicate, wherein reformatting the text phrase comprises excluding at least one of the plurality of words;

performing a task according to the reformatted text phrase; and

displaying a result associated with the task to a user.

12. The computer-readable medium ofclaim 11, wherein the text phrase comprises a conversion of a spoken phrase received from the user.

13. The computer-readable medium ofclaim 11, wherein the text phrase is associated with a web search query and identifying the predicate comprises defining a target domain.

14. The computer-readable medium ofclaim 13, wherein the reformatted text phrase comprises at least one keyword associated with the target domain.

15. The computer-readable medium ofclaim 14, wherein the target domain comprises a plurality of semantic slots.

16. The computer-readable medium ofclaim 11, further comprising:

filling at least one of the semantic slots with at least one slot word of the plurality of words.

17. The computer-readable medium ofclaim 16, wherein filling the at least one of the semantic slots comprises associating the at least one slot word of the plurality of words with the at least one of the semantic slots according to at least one second word of the plurality of words.

18. The computer-readable medium ofclaim 17, further comprising performing a dependency parse on the text phrase.

19. The computer-readable medium ofclaim 18, wherein the at least one second word comprises a dependent word associated with the at least one slot word during the dependency parse.

20. A system for providing sentence simplification, the system comprising:

a memory storage; and

a processing unit coupled to the memory storage, wherein the processing unit is operative to:

receive a spoken phrase from a user;

convert the spoken phrase to a text phrase, wherein the text phrase comprises a plurality of words,

perform a dependency parse on the text phrase,

identify a top level predicate of the text phrase,

reformat the text phrase according to the top level predicate, wherein the reformatted text phrase excludes at least one first word of the plurality of words,

define an intent domain according to the reformatted text phrase, wherein the defined intent comprises a plurality of syntactic slots,

fill at least one of the semantic slots with the at least one first word, wherein the at least one first word is associated with the at least one syntactic slot according to at least one dependent word of the at least one first word,

perform a task according to the defined intent and the filled at least one semantic slot, and

display a result of the performed task to the user.

Images