JP2011065401A - Information processor and data processing method and program thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information processor capable of efficiently generating an appropriate query, a data processing method and a program thereof, a document retrieval system, a call center operator support system, and a conference support system. <P>SOLUTION: The information processor includes; a query accumulation part 104 accumulating a query used for document retrieval; a reception part 110 receiving sentence data inputted occasionally; an extraction part 112 extracting a predetermined keyword when the received sentence data include the predetermined keyword; a generation part 114 referring to the query accumulation part 104 in each extraction of the keyword by the extraction part 112 and determining to which query the keyword is added based on a predetermined criterion to select a query to which the keyword is added or to generate a new query including the keyword; and an update part 116 accumulating the query to which the keyword is added or the generated new query in the query accumulation part 104. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an information processing apparatus, a data processing method thereof, and a program, and more particularly, to an information processing apparatus that can be used in a system that performs document search, a data processing method thereof, and a program.

  In order to search and present appropriate documents in a timely manner for utterances that are currently being made, such as telephone calls in contact centers, appropriate queries (keyword sets) corresponding to the situation during the call are generated at any time. It needs to be possible.

  As one method for generating such a query, a method of immediately estimating a topic during a call and performing a document search using a query corresponding to the topic can be considered. An example of a document search apparatus as a topic interval estimation system applicable to such document search is described in Patent Document 1.

  The document search apparatus of Patent Document 1 stores a predetermined keyword dictionary in advance, and stores keywords included in the keyword dictionary that appear in a recognition result for a call or the like in a contact center in the keyword history storage unit sequentially in time order. . Using the keyword list stored in the keyword history storage unit and the search database constructed from the search target document, the topic interval estimation processing unit sequentially estimates the topic interval during the call according to the call.

  Specifically, it is assumed that count values such as 1, 2, 3,... K are assigned in order from the oldest to the keywords stored in the keyword history in chronological order. Check whether there is a document in the search database that includes both keywords -1. Sequentially, the count value is decreased to k-2, k-3, and the operation is performed so that the point in time when there is no document including all of these keywords is used as the boundary of the topic section. At this time, the keyword set included in each topic section can be used as a query for document search.

  Further, the document search device described in Patent Document 2 is configured to generate a combination of keywords from a plurality of keywords stored in a keyword history extracted from a speech recognition result of conversation, for example, according to an extraction order. Have. At this time, it is described that one set of n keywords from the top or the end is selected, or that n keywords from the top are set as one set in descending order of the number of hits of the same keyword. Each time a new keyword is extracted during the conversation, the combination of keywords is changed. Further, the keyword combination can be changed by deleting a part of the keyword from the already existing keyword combination and adding a new keyword to the deleted keyword combination. In addition, keywords can be replaced or added using the synonym and synonym dictionary. In addition, it is possible to use a highly reliable keyword of the speech recognition result.

  The document search device described in Patent Document 3 detects a keyword from a character string being handled, accumulates it as a keyword appearance time series, accepts an input of a keyword directly related to the current response, and receives a keyword history. The related keyword relevant to the keyword input from is extracted, and the query is generated by combining the related keyword and the keyword. Thereby, it is possible not to select a keyword that is not directly related to the misrecognition or the topic of reception.

  In addition, the document search device described in Patent Document 4 has a configuration in which an undisplayed document title is presented from a search result that has been searched once, even in a situation where keywords are not extracted from the conversation. As a result, the amount of document information provided to the user is increased, and the target document can be reached efficiently.

  Further, in the information search device described in Patent Document 5, the search query construction module calculates the similarity between the query database that manages the query vector of the music that matches the user's preference and the query vector registered in the query database. A similarity calculation unit for calculating and a query vector integration unit for integrating a plurality of query vectors similar to each other are provided, and an accurate search is realized in a short time by applying a clustering method.

  The search system described in Patent Document 6 includes query keyword storage means for accumulating and storing keywords each time a keyword is input, and the keyword once used for the search is used as a new search keyword in a database or meta-data. Search efficiency can be improved by additionally registering data.

JP 2007-334429 A JP 2007-304793 A JP 2007-323558 A JP 2007-304776 A JP 2008-165714 A JP 2004-145706 A

In the document search system described in the above-mentioned patent document, when a keyword that is not actually spoken appears in the recognition result due to misrecognition, or when topic sections are nested or intersected Had the problem of not working properly.
The reason is that, for example, in the document search system described in Patent Document 1, a topic section is determined only by whether or not there is a document including a keyword list that is continuous in time, and if an incorrect keyword appears, that portion If the topic section is mistakenly divided or the topic sections are nested, the outer topic is divided by the inner topic section, and the document is searched using a query according to the incorrect topic section estimation result. It is because it will perform.

  In addition, the systems described in other patent documents have the same problems as described above because they are updated by adding keywords extracted to one query, replacing keywords, or deleting keywords.

  An object of the present invention is to provide an information processing apparatus, a data processing method, and a program for solving the above-described problem of generating an inappropriate query.

The information processing apparatus of the present invention
Query storage means for storing queries used for document search;
An accepting means for accepting text data input at any time;
When the received text data includes a predetermined keyword, an extraction means for extracting the keyword;
Each time the keyword is extracted by the extraction means, the query storage means is referred to, and it is determined whether or not to add the keyword to any query based on a predetermined criterion, and the query to which the keyword is added is selected. Or generating means for generating a new query including the keyword;
Updating means for storing the query to which the keyword is added or the generated new query in the query storage means.

The data processing method of the information processing apparatus of the present invention includes:
An information processing apparatus including a query storage device that stores queries used for document search is provided.
Accepts text data input from time to time,
When the received sentence data includes a predetermined keyword, the keyword is extracted,
Each time the keyword is extracted, the query storage device is referred to, it is determined whether or not to add the keyword to any query based on a predetermined criterion, and the query to add the keyword is selected or the keyword is selected. Generate a new query containing
The query to which the keyword is added or the generated new query is stored in the query storage device.

The program of the present invention
A program for realizing an information processing apparatus,
The information processing apparatus includes a query storage device that stores a query used for document search,
An acceptance procedure for accepting text data input from time to time;
An extraction procedure for extracting the keyword when the received text data includes a predetermined keyword;
Each time the keyword is extracted in the extraction procedure, the query storage device is referred to, it is determined whether or not the keyword is added to any query based on a predetermined criterion, and the query to which the keyword is added is selected. Or a generation procedure for generating a new query including the keyword,
An update procedure for storing the query to which the keyword is added or the generated new query in the query storage device is executed by a computer.

  The document search system of the present invention includes search means for performing the document search using the query stored in the query storage means of the information processing apparatus of the present invention.

  The call center operator support system of the present invention includes the document search system of the present invention.

  The conference support system of the present invention includes the document search system of the present invention.

  It should be noted that any combination of the above-described constituent elements and a conversion of the expression of the present invention between a method, an apparatus, a system, a recording medium, a computer program, etc. are also effective as an aspect of the present invention.

  The various components of the present invention do not necessarily have to be independent of each other. A plurality of components are formed as a single member, and a single component is formed of a plurality of members. It may be that a certain component is a part of another component, a part of a certain component overlaps with a part of another component, or the like.

  In addition, although a plurality of procedures are described in order in the data processing method and the computer program of the present invention, the described order does not limit the order in which the plurality of procedures are executed. For this reason, when implementing the data processing method and computer program of this invention, the order of the several procedure can be changed in the range which does not interfere in content.

  Furthermore, the plurality of procedures of the data processing method and the computer program of the present invention are not limited to being executed at different timings. For this reason, another procedure may occur during the execution of a certain procedure, or some or all of the execution timing of a certain procedure and the execution timing of another procedure may overlap.

  ADVANTAGE OF THE INVENTION According to this invention, the information processing apparatus which produces | generates a suitable query efficiently, its data processing method, a program, a document search system, a call center operator assistance system, and a meeting assistance system are provided.

It is a functional block diagram which shows the structure of the query production | generation apparatus which concerns on embodiment of this invention. It is a functional block diagram which shows the structure of the document search system which concerns on embodiment of this invention. It is a flowchart which shows an example of operation | movement of the information processing apparatus of this embodiment. It is a flowchart which shows an example of operation | movement of the query production | generation apparatus of this embodiment. It is a figure for demonstrating the difference of this invention and the system of patent document 1. FIG.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In all the drawings, the same reference numerals are given to the same components, and the description will be omitted as appropriate.

FIG. 1 is a functional block diagram showing the configuration of the information processing apparatus according to the embodiment of the present invention.
The information processing apparatus according to the present invention includes a query storage unit 104 that stores queries used for document search, a reception unit 110 that receives text data that is input as needed, and a text that includes predetermined keywords in the received text data. The extraction unit 112 that extracts the keyword, and each time the keyword is extracted by the extraction unit 112, the query storage unit 104 is referred to and it is determined based on a predetermined criterion whether the keyword is added to any query. A generation unit 114 that selects a query to add a keyword or generates a new query including the keyword, and an update unit 116 that accumulates the query to which the keyword is added or the generated new query in the query accumulation unit 104. Prepare.
In the present embodiment, the information processing apparatus includes a query generation apparatus 100 that generates a query used to search for a document to be searched.

  Each component of the information processing apparatus includes a CPU (Central Processing Unit) of an arbitrary computer, a memory, a program that realizes the components of this figure loaded in the memory, a storage unit such as a hard disk that stores the program, It is realized by any combination of hardware and software, centering on the network connection interface. It will be understood by those skilled in the art that there are various modifications to the implementation method and apparatus. Each figure described below shows functional unit blocks, not hardware unit configurations. In the following drawings, the configuration of parts not related to the essence of the present invention is omitted and is not shown.

  Specifically, the query generation device 100 includes a keyword dictionary storage unit (shown as “keyword dictionary” in the figure) 102, a query storage unit 104, a reception unit 110, an extraction unit 112, a generation unit 114, And an update unit 116. Further, the query generation device 100 is connected to the speech recognition processing unit 10.

Each of these units generally operates as follows.
The speech recognition processing unit 10 performs speech recognition processing on the input speech signal and outputs a recognition result to the query generation device 100. In the present embodiment, the voice recognition processing unit 10 inputs a voice signal of an utterance that is currently being made, such as a telephone call in a contact center.

  In the query generation device 100, the keyword dictionary storage unit 102 stores keywords used in document search, such as “database”, “Internet”, “browser”, “connection”, “cannot”, “phone”, “charge”. ”And the like are stored and prepared in advance, for example, based on a document to be searched. The query storage unit 104 stores a query used for document search. The keyword dictionary storage unit 102 and the query storage unit 104 are not necessarily included in the query generation device 100, and may be included in, for example, another storage device connected via a network. It only needs to be configured to read and write.

The accepting unit 110 accepts the recognition result output from the speech recognition processing unit 10.
The extraction unit 112 refers to the keyword dictionary storage unit 102 to check whether or not a keyword is included in the recognition result output from the speech recognition processing unit 10, and extracts the keyword if it is included. .

  In the present embodiment, the query is expressed by a keyword list, for example. For example, in the document search system 200 as shown in FIG. 2, the query stored in the query storage unit 104 is stored in the document database 210 by using the query stored in the query storage unit 104 of the query generation device 100. Can be used to search the document D1, D2, D3,... Stored in the table, and present the search result to the user at any time. The presentation unit 214 includes, for example, a display display unit (not shown), and can display the search result on the screen and present it to the user. The document search system 200 includes a search unit 212, a presentation unit 214, and a document database 210, and further includes a voice recognition processing unit 10 (FIG. 1) and a query generation device 100 (FIG. 1). Good.

  A plurality of queries are stored in the query storage unit 104 of the query generation device 100, and a document search can be performed for each of the queries. When the query is a keyword list, for example, a document search (so-called AND search) that outputs a document including all keywords in the keyword list can be performed. For example, a certain query is a query Q11 including a keyword such as “database, SQL” or a query Q12 including a keyword such as “Internet, connection”. The search unit 212 searches the document database 210 using these queries. For example, as a search result for the query Q12, the documents D2 and D3 are detected, and these documents are presented to the user by the presentation unit 214. Is done.

  Returning to FIG. 1, the generation unit 114 refers to the query stored in the query storage unit 104 and operates to update the query stored in the query storage unit 104 each time the keyword extracted by the extraction unit 112 is output. To do. For example, if the query is a keyword list, a query update can include adding a new keyword to one or more existing queries, or a query that consists of a new keyword alone. Including adding a new one. Of course, the new keyword may not be reflected in the query, that is, the query may not be updated.

  The generation unit 114 can use time information as a criterion for determining which existing query a new keyword is added to. That is, the query accumulating unit 104 stores the keyword in association with the keyword and the time information of the keyword, and the generation unit 114 compares the time of the latest keyword included in each query, and determines the newest keyword. New keywords can be added to the containing query.

  It is also possible to compare the time of the oldest keyword included in each query and add a new keyword to a query including the newest keyword among them. For each method, if the time difference between the keyword included in the query and the new keyword is equal to or greater than a predetermined threshold value, the new keyword can be not added to the query.

  In addition to the time information, the generation unit 114 can also use a relationship between keywords obtained from a search target as a determination criterion. For example, it is conceivable to add a new keyword to the query when the keyword included in the existing query and the new keyword are in the relationship using the relationship that the search objects appear simultaneously in one document. Further, when one or more search results are obtained by adding a new keyword to an existing query, it is possible to register the query.

  Furthermore, when the speech signal to be recognized is a dialogue by a plurality of speakers, the query can be updated using the response relationship of the extracted keywords. For example, in the case of an interaction between a contact center operator and a customer, for example, when an operator asks a plurality of questions and a customer responds to the question, a query is made by combining keywords having respective responsive relationships. Is to be generated. Of course, it is possible to combine these appropriately.

  Specifically, for example, when keywords having a responsive relationship are stored in the keyword dictionary storage unit 102 in association with each other in advance, when the keywords having the responsive relationship are extracted by the extracting unit 112, the generating unit 114 This can be determined by notifying this.

In this embodiment, the information processing apparatus including the query generation apparatus 100 includes, for example, a CPU, a memory, a hard disk, and a communication device (not shown), and is connected to an input device such as a keyboard and a mouse and an output device such as a display and a printer. It can be realized by a server computer, a personal computer, or a device corresponding to them. Each function of each unit can be realized by the CPU reading the program stored in the hard disk into the memory and executing it.
Further, the document search system 200 and the query generation device 100 described above may be included in the same device, or may be different devices.

  The computer program of the present embodiment includes a reception procedure for receiving sentence data that is input as needed by a computer for realizing the information processing apparatus (query generation apparatus 100), and a predetermined keyword is included in the received sentence data. When a keyword is extracted, the query storage unit 104 is referred to each time a keyword is extracted in the extraction procedure, and a query is used to determine whether a keyword is added to which query. A generation procedure for selecting a query to add a keyword or generating a new query including the keyword, and an update procedure for storing the query to which the keyword has been added or the generated new query in the query storage unit 104 It is described as follows.

  The computer program of this embodiment may be recorded on a computer-readable recording medium. The recording medium is not particularly limited, and various forms can be considered. The program may be loaded from a recording medium into a computer memory, or downloaded to a computer through a network and loaded into the memory.

The operation of the query generation device 100 of the present embodiment configured as described above will be described below.
3 and 4 are flowcharts illustrating an example of the operation of the query generation device 100 of the present embodiment.

  In the data processing method of the information processing apparatus (query generation apparatus 100) according to the present embodiment, the query generation apparatus 100 including the query accumulation unit 104 that accumulates queries used for document search accepts text data input as needed (see FIG. 3 (step S105), when a predetermined keyword is included in the received text data, the keyword is extracted (step S107 in FIG. 3), and each time a keyword is extracted, the query storage unit 104 is referred to. Whether a keyword is added to which query is determined based on a predetermined criterion (step S123 in FIG. 4), and a query to which a keyword is added is selected or a new query including the keyword is generated (step in FIG. 4). S127, S131), a query to which a keyword is added or a newly generated query is stored in the query storage unit 104. That (step S133 in FIG. 4).

FIG. 3 is a flowchart showing an example of the overall flow of operations of the query generation device 100 and the document search system 200 of this embodiment.
First, in the query generation device 100 of FIG. 1, the speech recognition processing unit 10 inputs a speech signal (step S101) and performs speech recognition processing (step S103). Then, the recognition unit 110 of the query generation device 100 receives the recognition result obtained by the speech recognition processing unit 10 (step S105), and the extraction unit 112 refers to the keyword dictionary storage unit 102 and includes a predetermined keyword. If so (YES in step S107), the extracted keyword is transferred to the generation unit 114, and the query generation process in FIG. 4 is called (step S109). If no keyword is included (NO in step S107), step S109 is bypassed and the process proceeds to step S111.

  Then, using the query updated in the query generation process and stored in the query generation device 100, the search unit 212 refers to the document database 210 and performs a document search in the document search system 200 of FIG. 2 (step S111). ). The presentation unit 214 presents the search result (step S113). It should be noted that a series of steps of this process is repeatedly executed until the voice input is completed or until a termination instruction is given by the user or the like. If no keyword is extracted in the determination in step S107 (NO in step S107), a document search using a query stored in the query storage unit 104 has already been performed in a series of conversations. If the result is presented and a new search result cannot be obtained, step S111 and step S113 may also be bypassed. That is, the process may return immediately to step S101 to prepare for the next keyword extraction.

  FIG. 4 is a flowchart illustrating an example of the flow of query generation processing when a keyword is extracted from the recognition result in the query generation device 100. That is, this process is called each time a keyword is extracted from the recognition result in step S107 in FIG.

  First, it is assumed that the generation unit 114 refers to the query storage unit 104 and repeatedly performs the following processing for all stored queries. For one of the queries stored in the query storage unit 104, the generation unit 114 determines whether or not to add a newly obtained keyword (step S123). This determination is made according to the determination criteria as described above. The user can select and specify which criterion is used for processing.

  Then, the process branches depending on whether or not a keyword is added to the query (step S125). That is, when a keyword is added to the query (YES in step S125), a new query in which a new keyword is added to the query is generated (step S127). When a keyword is not added to the query (NO in step S125), the generation unit 114 branches the process depending on whether there is a next query (step S129).

  That is, when there is a remaining query (NO in step S129), the process returns to step S123. On the other hand, if the processing is completed for all the queries and there are no remaining queries (YES in step S129), a new query including only the new keyword is generated (step S131).

  Finally, the update unit 116 reflects the query with the keyword added in step S127 and the new query generated in step S131 in the query storage unit 104 (step S133).

  The flow of processing described above is a procedure for updating one query for one keyword, but a plurality of queries may be updated. That is, for example, after adding a keyword to a query in step S127, the process may proceed to step S129 to repeat the determination of whether or not to add a keyword to all queries.

Further, it may be determined before step S131 whether or not to generate a new query consisting of a new keyword alone. The criteria for determining whether or not to generate a new query consisting of a new keyword alone is, for example, using the speech recognition score (reliability of word posterior probabilities, etc.) of the keyword as a criterion, and if the score is low, the query is By avoiding the generation, the influence of misrecognition can be reduced. This criterion can also be used in the same manner when determining whether or not to add a keyword to the query in step S125.
As described above, it is possible to perform a document search in the document search system 200 using a query that is generated by the query generation device 100 of the present embodiment and updated as needed based on the recognition result of the input voice signal. Become.

  Hereinafter, differences from the technique described in Patent Document 1 will be described with reference to FIG. As described above, the query generation device 100 (FIG. 1) according to the embodiment of the present invention refers to the queries and keywords stored in the query storage unit 104 (FIG. 1) according to the keywords extracted in time series. It is characterized by having a generation unit 114 (FIG. 1) and an update unit 116 (FIG. 1) that generate a simple query and store it in the query storage unit 104.

  As shown in FIG. 5A, in the technique shown in Patent Document 1, a search result is obtained in the document database 210 (FIG. 2) for a keyword string extracted in time series without having a query storage unit and a generation unit. Because it operates as a query with the largest (partial) keyword sequence that is included in time as included, it is ideal when topics are nested or when topics are mistakenly divided due to misrecognition. The query cannot be generated.

  For example, in FIG. 5A, two keywords “database” and “SQL”, which should originally be combined into one query, become two separate queries. On the other hand, in the present invention, as shown in FIG. 5B, two keywords “database” and “SQL” are combined into one query. In the present invention, instead of simply using a partial keyword sequence that is continuous in time as a query, each time a new keyword is extracted, the query stored at that time is referred to and joined to which query. In order to determine whether to use a single keyword or a new query, an appropriate query can be generated unlike the technique described in Patent Document 1.

Next, the effect of this embodiment will be described.
As described above, in this embodiment, the query generation device 100 includes the extraction unit 112, the query storage unit 104, the generation unit 114, and the update unit 116, and each time a keyword is obtained, the generation unit 114 allows the query to be updated using information such as the keyword appearance time interval and the relationship between speakers in a call as well as the time continuity of the appearing keyword. However, it is possible to provide the document search system 200 that can better cope with misrecognition and the appearance of various topic sections.

  As mentioned above, although embodiment of this invention was described with reference to drawings, these are the illustrations of this invention, Various structures other than the above are also employable.

  For example, a system that supports a call center operator can be realized using the document search system 200 of the above embodiment. In other words, when a call center operator has a conversation with a customer over the phone, it uses a query generated based on the keywords included in the result of recognizing the voice, and related to topics that appear in the conversation in a timely manner as needed. It is possible to search the document database 210 for information to be performed, for example, a relevant part of a manual, FAQ (Frequently Asked Questions), past similar response cases, etc., and present the search results.

  In addition, the document search system 200 of the above embodiment can be used for a conference support system. In other words, it recognizes the audio of the meeting and based on the keywords included in the recognition results, information related to topics that appear during the meeting in a timely manner as needed, for example, internal documents and related pages of the World Wide Web (WWW) Etc. can be presented.

  While the present invention has been described with reference to the embodiments and examples, the present invention is not limited to the above embodiments and examples. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

DESCRIPTION OF SYMBOLS 10 Speech recognition process part 100 Query generation apparatus 102 Keyword dictionary memory | storage part 104 Query storage part 110 Reception part 112 Extraction part 114 Generation part 116 Update part 200 Document search system 210 Document database 212 Search part 214 Presentation part

Claims (10)

Query storage means for storing queries used for document search;
An accepting means for accepting text data input at any time;
When the received text data includes a predetermined keyword, an extraction means for extracting the keyword;
Each time the keyword is extracted by the extraction means, the query storage means is referred to, and it is determined whether or not to add the keyword to any query based on a predetermined criterion, and the query to which the keyword is added is selected. Or generating means for generating a new query including the keyword;
An information processing apparatus comprising: an update unit that stores the query to which the keyword is added or the generated new query in the query storage unit. The information processing apparatus according to claim 1,
The text data received by the receiving means is an information processing apparatus which performs voice recognition processing on an input voice signal and is recognition result data thereof. The information processing apparatus according to claim 1 or 2,
The query storage means stores the keyword and time information of the keyword in association with each other,
The generating unit is an information processing apparatus that uses the time information as the determination criterion. The information processing apparatus according to claim 1,
The information processing apparatus uses the relationship between the keywords as the determination criterion. The information processing apparatus according to claim 4,
The information processing apparatus uses the responsive relationship between the keywords as the determination criterion. An information processing apparatus including a query storage device that stores queries used for document search is provided.
Accepts text data input from time to time,
When the received text data includes a predetermined keyword, the keyword is extracted,
Each time the keyword is extracted, the query storage device is referred to, and it is determined based on a predetermined criterion whether or not the keyword is added to any query, and the query to which the keyword is added is selected or the keyword is selected. Generate a new query containing
A data processing method of an information processing apparatus for storing the query to which the keyword is added or the generated new query in the query storage apparatus. A program for realizing an information processing apparatus,
The information processing apparatus includes a query storage device that stores a query used for document search,
An acceptance procedure for accepting text data input from time to time;
An extraction procedure for extracting the keyword when the received text data includes a predetermined keyword;
Each time the keyword is extracted in the extraction procedure, the query storage device is referred to, it is determined whether or not the keyword is added to any query based on a predetermined criterion, and the query to which the keyword is added is selected. Or a generation procedure for generating a new query including the keyword,
A program for causing a computer to execute an update procedure for storing the query to which the keyword is added or the generated new query in the query storage device.   A document search system comprising search means for performing the document search using the query stored in the query storage means of the information processing apparatus according to claim 1.   A call center operator support system including the document search system according to claim 8.   A conference support system including the document search system according to claim 8.

Images