JPH07295994A - Information retrieval device - Google Patents

Abstract

PURPOSE:To omit a waste such that the user observes a retrieval result of the same content for plural number of times by deleting other results while leaving only one result among same content results when results of the same contents are included in results of plural retrieval acts. CONSTITUTION:A receiver terminal equipment uses a retrieval section 4 to make retrieval in response to a retrieval request entry, the retrieval result is stored tentatively in a retrieval result buffer 5, and the reception terminal equipment sends the retrieval result to a sender terminal equipment via a network section 2. The sender terminal equipment uses a duplication delete section 6 to check whether or not results of same contents are included among plural retrieval results and deletes other results while leaving one result of the same contents when the same results are included in them. Thus, the retrieval results of duplicate contents are not outputted, and then waste of the user observing the retrieval results of same contents for plural number of times is avoided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieving apparatus such as an electronic notebook, a portable information terminal, and a word processor connected by a network.

[0002]

2. Description of the Related Art Conventionally, when searching information stored in a device connected to a network, search results are displayed in order of similarity to a given search expression, order of search, or order of creation date. There is. For example, JP-A-5-108726
Japanese Patent Publication No. JP-A-2003-242242 discloses a search device that performs a flexible search process by calculating the similarity between a keyword and an index over a plurality of thesauri in response to an ambiguous search keyword, and outputs the search results in descending order of similarity. Is disclosed.

Further, since the conventional information retrieval apparatus has a database created assuming a large number of users and a small number of administrations, if there is no retrieval result, only a reply that the retrieval result cannot be found is returned. come.

Conventionally, the keyword search is most often used in the search. In this case, the search is performed by partially matching or completely matching the input keyword with the search target. A method of extracting a keyword related to a keyword included in an input search request from a thesaurus and expanding a search expression is disclosed in Japanese Patent Laid-Open No. 5-250411.
It is described in Japanese Patent Publication No. Further, in a semantic search or an associative search in which a search is performed in consideration of the content of a document, a document information search apparatus that obtains a similarity between a keyword and document content and uses a result having a similarity close to the search result is known. -281
No. 98 publication.

[0005]

However, in the search device described in Japanese Patent Laid-Open No. 5-108726, even if the same search result is included, all the results are presented to the user, so the user can see the contents. For the first time, it turns out that the same search results are obtained, and thus unnecessary operations are performed. In particular,
In the situation where a search request is issued to multiple terminals connected via a network, there is no problem that the same search results are displayed in duplicate when operating as one system as a distributed database. If the information terminals are connected to the network and individual data can be retrieved from each terminal, there is a high possibility that each terminal stores the same information, so the same search result will be duplicated. May be displayed.

For example, Mr. A creates a document a relating to image processing. Mr. B searches the documents related to image processing from the data of Mr. A and Mr. C. As a result, the document a is copied to Mr. B's information terminal. Mr. C searches for documents related to image processing from the data of Mr. A and Mr. B. The document a owned by Mr. A and the document a owned by Mr. B are copied and presented to Mr. C. As a result, Mr. C sees two identical documents a.

Even if there is no overlap, information that is not very related will be displayed in succession in the order of similarity of search, and it will be very difficult to see if there are a large amount of search results.

[0008] Recently, the database has been downsized so that an individual can carry it around, and if the terminal is connected to the network at the destination, a problem that cannot be dealt with arises. For example, suppose that the user asks the location of the tomorrow's confirmation meeting to a terminal connected to the network. If it is not registered in the database, the user again asks someone who is likely to know the information. When the portable information tool also serves as a database, once a search request is issued, and if the response is not obtained, then another information tool connected to the network by communication means such as e-mail is used. You have to ask the user a question, and you have to ask the same question to both the database and the person.
The trouble is complicated.

It is preferable that the keyword search and the meaning search are used properly according to the usage. For example, when it is desired to obtain information on a company called "applied system research and development company", this search request is a proper noun, so it is preferable to perform keyword matching. It is useless to disassemble "application system research and development company" and search for information close to the meanings of the words "application", "system", "research", "development", "stock", and "company". In the search request of “image processing chip”, not only the words “image”, “processing”, and “chip” themselves match, but also similar terms such as “image” and “LSI” are searched. desirable. Depending on the object to be searched as described above, there are cases where the keywords need to be completely matched and cases where it is better to search for information that is semantically similar. JP-A-5-250411 and JP-A-5-250411 −
In the conventional document information search device described in Japanese Patent No. 28198, these cannot be used in combination, and if there is no corresponding item in the keyword search, the meaning search is performed again with the same keyword. This has to be done and the effort is complicated. When there are a large number of relevant items in the semantic search, it is necessary to search for the matching items by keyword matching from the information retrieved by the semantic search, which is troublesome.

The present invention has been made in order to solve the above problems, and a first object of the present invention is to obtain the same contents when a plurality of search results include the same contents. An object of the present invention is to provide an information retrieval device that can leave only one result and delete other results. A second object is to provide an information retrieval device that classifies a plurality of retrieved results according to their contents. A third object is to provide an information search device that requests an answer from a user when there is no answer to the search request. A fourth object is to provide an information search device capable of selectively using keyword search and meaning search.

[0011]

According to the present invention, the above first object is to provide an input means for inputting search data, a data storage means for holding searched data, and a search for a search request. Search means for performing, buffer means for temporarily holding the searched search results, display means for displaying the searched data and the search results, the search request, and a network for communicating the search results via a network Means, and if a plurality of the above-mentioned search results include the same search result, and if the same search result is included, only one search result of the same search results The information retrieval apparatus according to claim 1, further comprising: a deleting unit that deletes other search results that remain.

According to the present invention, the above-mentioned first object is that the deleting means assigns an ID given to the data at the time of creating the data.
This is achieved by the information retrieval device according to claim 2, further comprising a determination unit that determines that the data are the same based on the number.

According to the present invention, the aforementioned second object is to provide an input means for inputting search data, a data storage means for holding the searched data, a search means for performing a search in response to a search request, and a search. Buffer means for temporarily holding the retrieved search result, display means for displaying the searched data and the retrieval result, the retrieval request, and a network means for communicating the retrieval result via a network, It is achieved by the information search device according to claim 3, further comprising a classification unit that classifies a plurality of the search results.

According to the present invention, the above-mentioned third object is to provide an input means for inputting search data, a specifying means for specifying another information search device on the network to be searched, and the specified other information. First transmitting means for transmitting a search request to the search device, first receiving means for receiving the search request via the network, data storage means for storing search data, and based on the search request Searching means for searching with reference to the data storage means, buffer means for holding the search results, second transmitting means for sending the search results through the network, and receiving the search results through the network Second receiving means, display means for displaying the search result, and when there is no search result, presents the search request to the user and returns a response from the user to the net. It is achieved by the information retrieval apparatus according to claim 4 and a question means for receiving through over click.

According to the present invention, the above-mentioned third object is that the inquiring means determines the result of the search whether there is the search result and the search request when there is no search result. The information retrieval apparatus according to claim 5, further comprising request display means for presenting to the user, answer input means for inputting the user's answer to the presented search request, and answer transmitting means for transmitting the entered answer. It

According to the present invention, the above third object is that the information retrieval apparatus further comprises a mode setting means for setting a user inquiry mode as to whether or not to ask a question to the user, and the user inquiry mode. The information retrieval apparatus according to claim 6, further comprising: a mode buffer for storing the information, and a mode determination means for determining the user inquiry mode.

According to the present invention, the above-mentioned fourth object is achieved by input means for inputting a search request, interpreting means for interpreting the search request, data storage means for storing searched data, and keyword matching. The information search device according to claim 7, further comprising: a keyword search means for performing a search, a meaning search means for performing a search according to meaning, a search result buffer means for temporarily holding a search result, and a display means for displaying the search result. To be achieved.

According to the present invention, the above-mentioned fourth object is obtained by the meaning retrieving means, a document analyzing means for analyzing a stored document to extract an independent word, and the document analyzing means. A word vector generation means for creating a word vector by reading the feature amount corresponding to the independent word from the feature amount dictionary,
A word vector storage unit that stores the generated feature amount of the independent word, and a sentence vector that represents the feature of the sentence is obtained and stored using one or more of the word vectors stored in the word vector storage unit. 9. The sentence vector storage means, and the similarity extraction means for obtaining the similarity between the feature quantity of the word included in the search request and the feature quantity of the word vector stored in the word vector storage means.
It is achieved by the information retrieval device of.

According to the present invention, the above-mentioned fourth object is that the information retrieval apparatus further comprises result determining means for determining the retrieval result and the retrieval request when the retrieval result is not satisfied. It is achieved by the information retrieval device according to claim 9, which comprises a correction means.

[0020]

According to the information retrieving apparatus of the present invention, in the transmitting side terminal, the contents of retrieval / question are inputted by the input means, and the retrieval request is communicated to the plurality of receiving side terminals via the network by the network means. It In the receiving side terminal, the search means performs a search for the search request input, and the search result is temporarily held by the buffer means.
The search result is transmitted from the receiving side terminal to the transmitting side terminal via the network. In the sending terminal, the deleting means checks whether or not the same content is included in multiple search results, and if the same result is included, one result of the same content is found. Other results are deleted, leaving only. Therefore, the search result of the duplicated content is not output, and it is possible to omit the waste that the user views the search result of the same content multiple times.

According to the information retrieving device of the second aspect, in the transmitting side terminal, the contents of retrieval / question are inputted by the input means, and the retrieval request is communicated to the plurality of receiving side terminals via the network by the network means. It In the receiving side terminal, the search means performs a search for the search request input, the search result is temporarily held by the buffer means, and the search result is transmitted from the receiving side terminal to the transmitting side terminal via the network. . In the transmitting terminal, the deleting means checks whether or not the plurality of search results include the result of the same ID number. If the result of the same ID number is included, the result of the same ID number is deleted. Only the result of one ID number is left, and the results of other ID numbers are deleted. Therefore, the search result of the duplicated content is not output, and it is possible to omit the waste that the user views the search result of the same content multiple times.

According to the information retrieving device of the third aspect, in the transmitting side terminal, the contents of retrieval / inquiry are inputted by the input means, and the retrieval request is communicated to the plurality of receiving side terminals via the network by the network means. It In the receiving side terminal, the search means performs a search for the search request input, the search result is temporarily held in the search result buffer means, and the search result is transmitted from the receiving side terminal to the transmitting side terminal via the network. To be done. The classifying unit classifies the plurality of retrieved results according to the contents. Therefore, since information having a meaningful cohesion is continuously presented, the user can easily understand the search result.

According to the information retrieving apparatus of claim 4, in the transmitting side terminal device, the contents of retrieval / question are inputted by the input means, and the receiving side terminal device to be retrieved is designated by the designating means, and the first transmitting means. The search request is transmitted to the other receiving terminal specified by. On the receiving terminal,
The search request receiving means receives the search request via the network, the search means performs the search by referring to the data storage means based on the search request, the buffer means holds the search result, and the search result determining means searches the search result. It is determined whether or not there is, and if there is a search result, the search reply is transmitted through the network, and the search reply is received by the second receiving means. If there is no search result, the inquiry means presents a search request to the user, inputs the user's answer to the presented search request, and transmits the input answer to the transmitting terminal. Therefore, when there is no reply to the search request, the user is automatically requested to reply, and the trouble of replying to the user again can be omitted.

According to the information retrieving apparatus of claim 5, in the transmitting side terminal device, the contents of retrieval / question are input by the input means, the receiving side terminal device for performing the retrieval is designated by the designating means, and the retrieval request transmitting means is constructed. A search request is sent to the other designated receiving terminal. In the receiving side terminal, the search request is received by the first receiving means via the network, the search means searches the data storage means based on the search request, and the buffer means holds the search result. The search result determining means determines whether or not there is a search result. If there is a search result, the search response is transmitted through the network, and the second receiving means receives the search response. If there are no search results,
The search means is presented to the user by the display means, the user's answer to the search request presented by the answer input means is input, and the answer entered by the answer sending means is transmitted to the sending terminal. Therefore, when there is no reply to the search request, the user is automatically requested to reply, and the trouble of replying to the user again can be omitted.

According to the information retrieving device of the sixth aspect, in the transmitting side terminal, the contents of retrieval / question are inputted by the input means, and the receiving side terminal to be retrieved is designated by the designating means, and the first transmitting means. The search request is transmitted to the other receiving terminal specified by. On the receiving terminal,
The first receiving means receives the search request via the network, the search means searches the data storage means based on the search request, the search result buffer means holds the search result, and the search result determining means. Determines whether or not there is a search result. If there is a search result, the search reply is transmitted through the network, and the search reply is received by the second receiving means. There is no search result, and the user question mode set by the mode setting means is'YE
When S'is determined by the mode determination means, the user inquiry means presents the search request to the user, the user's answer to the presented search request is input, and the input answer is transmitted to the transmitting terminal. .

Therefore, if the user of the receiving terminal does not want to answer a question from another user, the user question mode can be set to'NO 'without answering the question from another person. For example, when the user of the receiving terminal is very busy, the answer is "NO" so that the question is not presented, and when the user is not busy, the answer is "YES" to answer the question. Further, when the user rarely uses the receiving terminal, even if the user asks a question, the opportunity for the user to touch the question is limited, so the mode may be set to “NO”. For example, when the receiving terminal is performing reservation management work in a conference room, there is no specific user who directly operates the terminal, but rather a large number of unspecified information inputs and inquiries. Thus, the device alone manages the allocation of conference rooms and the information related thereto. Even in such a case, since the person who answers the question to the receiving terminal is unknown, set the mode to'NO 'and prohibit the question to the user of the receiving terminal to prevent confusion. You can

According to the information retrieving apparatus of claim 7, the retrieval request is inputted by the input means, the retrieval request is interpreted by the interpreting means, and the keyword retrieval means carries out the keyword matching from the retrieval target data stored in the data storage means. A search is performed, the meaning search means performs a search according to the meaning and content, the search result buffer means temporarily holds the search result, and the display means displays the search result.
As a result, it is possible to selectively use the keyword search and the semantic search with a single search request, and it is possible to cope with a wider search request.

According to the information retrieval apparatus of the eighth aspect, the semantic retrieval is performed as follows. The document analysis means analyzes the stored document and extracts an independent word. The word vector generation means reads the feature quantity corresponding to the independent word obtained by the document analysis means from the feature quantity dictionary to create a word vector. The word vector storage unit stores the generated feature amount of the independent word. The sentence vector storage means obtains and stores a sentence vector representing a feature of a sentence using one or more word vectors stored in the word vector storage means. The similarity extraction unit obtains the similarity between the feature amount of the word included in the search request and the feature amount of the word vector stored in the word vector storage unit. Therefore, a semantic search can be performed.

According to the information retrieving apparatus of the ninth aspect, the retrieval request is inputted by the inputting means, the retrieval request is interpreted by the interpreting means, and the retrieval data stored in the data storing means is subjected to keyword matching by the keyword retrieving means. A search is performed, the meaning search means performs a search according to the meaning and content, the search result buffer means temporarily holds the search result, and the display means displays the search result.
The result determination means determines the search result. If the search result is not satisfactory, the correction means corrects the search request. Further, a search is performed. Therefore, it is possible to perform a search that makes the best use of the advantages of both the keyword search and the semantic search.

[0030]

EXAMPLE An example of the information retrieval apparatus according to claim 1 of the present invention will be described below. The present embodiment has an object to provide an information search device capable of deleting one result while leaving only one result having the same content when a result having the same content is included in a plurality of search results. To do.

In the information retrieval apparatus of this embodiment, a plurality of terminals are connected to each other via a network. As shown in FIG. 1, the terminal device includes an input unit 1 as an input unit for inputting search / question contents, a network unit 2 as a network unit for communicating a search request and a search result via a network, and a search target. A data storage unit 3 as a data storage unit for holding data, a search unit 4 as a search unit for performing a search in response to a search request input, a search result buffer 5 as a buffer unit for temporarily holding a search result,
If multiple search results obtained from other terminals include the same result, and if the same result is included, only one result of the same content is found. And a display unit 7 as a display unit for displaying the input status, the search status, the search result, and the like.

For example, assume a case of inquiring about a company's schedule for the next month using a portable information terminal. The mobile information terminal issues a search request to another person's mobile information terminal connected to the network. When a search request is issued to a plurality of terminals, two or more terminals may have the same information among them. In such cases,
The same result is sent as a search result from different terminals. When the information is displayed on the user's terminal without performing any processing, a plurality of pieces of information having the same content are displayed, so that the confirmation work is duplicated and extra time is required. Therefore, by deleting the duplicated information as in the present invention, the confirmation time can be shortened.

The operation of this embodiment will be described below with reference to the flow charts of FIGS. 2a and 2b.

At the sending terminal, the user waits for the search conditions to be input (step S
1) When a search request is input by the input unit 1, the network unit 2 interprets the search request and sends it to another terminal connected to the network (step S2).
Waiting for an answer from another terminal (step S
3). When one search answer is received by the network unit 2, the duplication eliminator 6 determines whether or not it matches with the answers received so far (step S4). If the received answer matches the already received answer, one of the duplicated data is deleted by the duplication eliminator 6 (step S5). It is determined whether all the answers have been returned (step S6), and the display unit 7 displays the result excluding the overlapping portion (step S7). Even before all the answers are returned, according to the instructions from the user,
The reception result at the current stage may be displayed.

Various methods are conceivable as a method for determining duplicated data. For example, an ID number is given at the time of data creation, and the ID number is taken over at the time of copying. If corrected, a new ID is given. in this way,
The ID numbers are compared, and if there is a matching ID, it is deleted. I
If there is no D, a method of comparing the texts and erasing if all the texts match can be considered.

When the terminal on the receiving side receives the search request from the terminal on the transmitting side (step S8), the retrieval unit 4 holds the data stored in the data storage unit 3 in accordance with the search request from the terminal on the transmitting side. The search data is searched (step S9). The retrieved data is temporarily stored in the retrieval result buffer 5 and then transmitted by the network unit 2 to the transmitting terminal via the network (step S10).

Therefore, the search results having the duplicated contents are not output, and it is possible to omit the waste of the user viewing the search results having the same contents a plurality of times.

An embodiment of the information retrieval apparatus according to claim 3 of the present invention will be described below. An object of the present embodiment is to provide an information retrieval device that classifies a plurality of retrieved results according to their contents.

The terminal of the information retrieval apparatus of this embodiment is shown in FIG.
Input section 1 for inputting search / question content, as shown in
A network unit 2 that communicates a search request and a search result via a network, a data storage unit 3 that holds search target data, and a search unit 4 that searches for a search request input.
A search result buffer 5 that temporarily holds search results, a result classification unit 8 that classifies search results by content, and a display unit 7 that serves as a display unit that displays input status, search status, search results, and the like.
It has and. As the classification method by the result classification unit 8, for example, the classification method by the document classification device described in Japanese Patent Application No. 5-259809 already proposed by the present applicant is used.

As shown in FIG. 4, the result classification section 8 analyzes the document of the search response and extracts the independent word, and the document analysis section 9
And a word vector generation unit 11 that reads a word vector for an independent word registered in the vector generation dictionary 10 among the independent words obtained by the document analysis unit 9, and temporarily stores the read word vector. By comparing the word vector storage unit 12, the sentence vector storage unit 13 that obtains a sentence vector from all the word vectors obtained in all sentences, and the sentence vector comrades obtained in the sentence vector storage unit 13, A classification unit 14 that obtains a distance and performs classification using the distance, and a result storage unit 15 that temporarily stores the classification result are provided. For classification, it is possible to use a K-means method or a neural network method such as a self-organizing map as a statistical method.

Search answers classified into a plurality of groups are
It is also possible to display in each classification group together with a keyword representing that group. The detection of the keyword can be realized by using the keyword extracting method described in, for example, Japanese Patent Laid-Open No. 5-274351.

Next, the operation of this embodiment will be described with reference to the flowchart of FIG.

At the sending terminal, the user waits for the user to input search conditions (step S1).
1) When the input unit 1 inputs a search request such as a keyword to be searched and a search target device name as shown in FIG. 6,
The network unit 2 interprets the search request and transmits it to another terminal connected to the network (step S
12). It waits for a reply from another terminal (step S13). It is determined whether or not all the answers are returned (step S14), and when it is determined that all the answers are returned, the result classification unit 8 classifies the search results by content (step S15). The display unit 7 displays the classification result (step S17). As a result, semantically coherent information is continuously presented, and the user can easily understand the search result.

For example, the search result is as shown in FIG.
It is classified into three groups, and keywords such as "system", "specification, performance", and "development and sale" are given to each group. By specifying the group that is most interesting to the user, the user can see the search answers related to it. Of course, in addition to the mode in which the classification is performed automatically, the classification is performed by the user's designation, and other display methods such as the order of similarity and the order of creation date can be selected.

An embodiment of the information retrieval apparatus according to claim 4 of the present invention will be described below. The present embodiment has an object to provide an information search device that requests an answer from a user when there is no answer to the search request.

In the information retrieval apparatus of this embodiment, a plurality of terminals are connected to each other via a network. As shown in FIG. 8, the terminal unit, as an input unit for inputting search / question contents, and an input unit 16 for searching, searches for data only for the terminal unit or connects to the network. And a search request as a first transmitting means for transmitting a search request to the specified receiving side terminal as a specifying means for specifying whether to perform a search for the other specified terminal. A transmission unit 18, a search request receiving unit 19 as a first receiving unit that receives a search request via a network, a data storage unit 20 as a data storage unit that stores search data, and data storage based on a search request. A search unit 21 as a search unit that performs a search by referring to the unit 20, a search result buffer 22 as a buffer unit that temporarily holds a search result, and a search result through a network. Find a second receiving means result receiving unit 2
3, a search result transmitting unit 24 for transmitting the search result through the network, a display unit 25 as a display unit for displaying the search result, and a search result determining unit as a result determining unit for determining whether or not there is a search result. 26, a search request display unit 27 as a request display means for presenting a search request to the user when there is no search result, and a user answer input means as a user answer input means for inputting the user's answer to the presented search request. It is provided with a unit 28 and a user response transmitting unit 29 as a response transmitting means for transmitting the input response.

Next, the operation of this embodiment will be described with reference to the flowchart of FIG.

It is determined whether or not the search request is input from the input unit 16 of the terminal itself (step S21), and if the search request is not input from the input unit 16 of the terminal itself,
It is determined whether a search request has been input from another terminal via the network (step S22). When a search request is input from another terminal, the search is performed by the search unit 21 based on the search request (step S23).
The search result determination unit 26 determines whether or not there is a search result (step S24). If there is a search result, the search result transmission unit 24 sends the search result to the transmission side terminal device via the network (step S25). ). If there is no search result, the search request display unit 27 displays the search request input via the network (step S2).
6) The user's reply to the displayed search request is awaited (step S27). User answer input section 28
When the answer is input from the user, the answer is transmitted by the user response transmitting unit 29 to the transmitting side terminal that has issued the search request (step S28). Step S21
In, when the search request is input from the input unit 16 of the terminal itself, it is selected whether to perform the search only in the terminal or to issue the search request to another terminal connected to the network ( Step S29). When the search is selected only within the terminal, the search is performed by the search unit 21 (step S30), and the search result is displayed on the display unit 25. When it is selected to issue a search request to another terminal connected to the network, the search request transmission unit 18 transmits the search request to the other terminal via the network (step S3).
1).

A search is performed by the search unit of a plurality of terminals connected to the network that receives the search request (step S32), and the search result is transmitted. This search result is received by the search result receiving unit 23 (step S3
3), the received search result is temporarily stored in the search result buffer 22 and then displayed on the display unit 25 (step S3).
4).

For example, a case where Mr. X inputs a search request as shown in FIG. 10 will be described.

FIG. 11 is displayed on the display unit 25 of Mr. X's terminal.
As shown in FIG. 5, “About Company Z image processing chip” is displayed as the search content, and “A's terminal and B's terminal” connected to the network are displayed as designated devices to be searched. Then, on the display unit of Mr. B's terminal, as shown in FIG. 12, it is displayed that Mr. X has made a search request "about image processing chip of company Z", and there is no item corresponding to the search request. So, the question to Mr. B himself is automatically displayed. When this display is confirmed by Mr. B, Mr. B inputs the response that "the sample price of the Z company image processing chip is 5000 yen, and the shipment will start from December." When the input is completed, B
The answer is sent from Mr. X's terminal to Mr. X's terminal via the network. In this series of operations, Mr. X only needs to enter a search request once, and the terminal connected to the network and the user using it, in this case B
A search has been performed for both and.

Therefore, when there is no reply to the search request, the user is automatically requested to reply, and it is possible to omit the trouble of replying to the user again.

An embodiment of the information retrieval apparatus according to claim 6 of the present invention will be described below with reference to FIG. In this example,
An object of the present invention is to provide an information search device capable of designating whether or not to ask a question to the user of the terminal when there is no search result. The same components as those in FIG. 8 are designated by the same reference numerals and the description thereof will be omitted.

The information retrieving apparatus of the present embodiment sets whether or not to ask the user of the terminal when the retrieval fails when the retrieval request is made to another terminal connected to the network. The user question mode setting unit 30 to perform, the user question mode buffer 31 that temporarily stores the user question mode associated with the received search request, and the case where there is no corresponding data in the search request transmitted via the network And a user question mode determination unit 32 for determining whether to ask the user of the terminal by using the user question mode.

For example, consider the case where Mr. X asks a certain question to Mr. A's terminal and Mr. B's terminal.
If the question cannot be answered by Mr. A's terminal and Mr. B's terminal, the user can set a user question mode for directly asking Mr. A or Mr. B. In this case, if the answer to the question cannot be obtained from Mr. A's terminal, the question is asked to Mr. A. As a result, Mr. X will get A if his terminal has an answer.
You can get an answer from Mr. A's terminal, and if Mr. A's terminal does not have an answer, you can get an answer from Mr. A himself. If Mr. A's terminal does not have an answer, it is possible to make settings so that A is not asked.

Next, the operation of this embodiment will be described with reference to the flowchart of FIG. It should be noted that the same steps as those in FIG.

If it is determined in step S24 that there is no search result, the user inquiry mode determination unit 32 determines whether the search request input via the network is in the user inquiry mode (step S35), and the user inquiry is asked. If it is not in the mode, the search reply transmission unit 24 transmits the search failure to the transmission side terminal (step S36). In the user inquiry mode, the operations after step S26 are performed.

When it is selected in step S29 to issue a search request to another terminal connected to the network, when the user inquiry mode setting unit 30 issues a search request to another terminal. A mode for asking a question to the user of the terminal when no search result is obtained is selected (step S37), and a user question mode is set (step S38).

If the search fails on the terminal side,
Although the situation of asking the user a question has been explained, the present invention is not limited to this, and even if the search is successful on the terminal side, it is possible to present a search request to the user and obtain both the search result and the user's answer. It is possible.

[0060] Further, the search is not limited to the narrow sense of obtaining matching information by matching with stored data, but also "Please tell me the shortest route from Tokyo station to Shinjuku station". It may be the content of a question that can be answered by various artificial intelligence technologies.

Therefore, if the user of the receiving terminal does not want to answer a question from another user, the user question mode can be set to "NO" so that the question from another person is not answered. For example, when the user of the receiving terminal is very busy, the answer is "NO" so that the question is not presented, and when the user is not busy, the answer is "YES" to answer the question. Further, when the user rarely uses the receiving terminal, even if the user asks a question, the opportunity for the user to touch the question is limited, so the mode may be set to “NO”. For example, when the receiving terminal is performing reservation management work in a conference room, there is no specific user who directly operates the terminal, but rather a large number of unspecified information inputs and inquiries. Thus, the device alone manages the allocation of conference rooms and the information related thereto. Even in such a case, since the person who answers the question to the receiving terminal is unknown, set the mode to'NO 'and prohibit the question to the user of the receiving terminal to prevent confusion. You can

An embodiment of the information retrieval apparatus according to claim 7 of the present invention will be described below. The present embodiment has an object to provide an information search device capable of selectively using a keyword search and a semantic search.

As shown in FIG. 15, the information retrieving apparatus of this embodiment has an input unit 3 as an input means for inputting a retrieval request.
3, a search request interpretation unit 34 as an interpretation unit that interprets a search request, a search range setting unit 35 that sets a range of data to be searched, and a data storage unit 36 that is a data storage unit that stores the searched data. , A keyword search section 37 as a keyword search means for performing a search by keyword matching, a meaning search section 38 as a meaning search means for performing a search according to meaning and content, and a search result as a buffer means for temporarily holding a search result. A buffer 39 and a display unit 40 as a display unit for displaying the search result are provided.

As shown in detail in FIG. 16, the meaning retrieving unit 38 analyzes a stored document to extract an independent word, and an independent word obtained by the document analyzing unit 41. A word vector generation unit 43 that reads a corresponding feature amount from the feature amount dictionary 42 to create a word vector, and a word vector storage unit 44 that stores the generated feature amount of the independent word.
And a sentence vector storage unit 45 that obtains and stores a sentence vector that represents a feature of a sentence using one or more word vectors stored in the word vector storage unit 44, and a similarity degree extraction unit 46 that extracts a degree of similarity. It has and. For example, the sentence vector storage unit 45 is configured to obtain an average of the feature amounts of the word vectors stored in the word vector storage unit 44, and the similarity extraction unit 46 causes the feature amount of the words included in the search request. And the similarity with the feature amount of the word vector stored in the word vector storage unit 44.

By the processing blocks 41 to 45, the feature amount of the sentence is obtained, and the similarity between the feature amount and the feature amount of the word of the search request is obtained. From all the sentences to be searched, some of the sentences with the highest similarity may be set as the search result. The details of this processing are described in Japanese Patent Application No. 5-259809 filed by the applicant of the present invention, and therefore the description thereof is omitted.

The feature quantity dictionary 42, as shown in FIG.
A document storage unit 47 that stores a large number of documents, and a sentence analysis dictionary 4
8, a sentence analysis unit 49 that extracts an independent word, an independent word storage unit 50 that stores an independent word obtained from a sentence, a co-occurrence relationship creation unit 51 that creates a co-occurrence relationship, and a co-occurrence relationship A co-occurrence registration dictionary 52 for registration and a co-occurrence relation storage unit 53 for storing a co-occurrence relation table obtained by performing the co-occurrence relation creating unit 51 on all sentences are provided. The obtained co-occurrence relation is used as a feature amount. The texts stored in the document storage unit 4 are texts for creating the feature quantity dictionary and may not be related to the search. If the word A and the word B appear in the same sentence, the co-occurrence relation creating unit 51 determines that the word A and the word B are included.
It is considered that there is some relationship with and, and an operation of adding 1 to the element of the word B of the feature vector of the word A is performed. The means for creating the characteristic amount is described in Japanese Patent Application No. 5-181264 filed by the applicant of the present invention, and therefore its explanation is omitted.

Next, the operation of this embodiment will be described with reference to the flowchart of FIG.

The input of the search condition is awaited by the user (step S41), and when the search request is input, the search request interpreter 34 interprets the search request (step S42). For example, if the form of the search expression is input, the AND and OR searches are interpreted. The search range setting unit 35 sets the search range (step S43). For example, the range of creation time of data to be searched based on the search request is limited, and
If it is a search, the next search is set only from the data already searched and stored.
It is determined whether the current search element is a keyword search or a semantic search is specified (step S4).
4). When the keyword search is designated, the keyword search unit 37 performs the keyword search (step S45). When the semantic search is designated, the semantic search unit 38 performs the semantic search (step S46).
The search result is stored in the search result buffer 39 (step S47). It is determined whether or not the search request is over (step S48), and if it is over, the display unit 40 displays the search result.

As shown in FIG. 19, the input example of the search condition requires the user to clearly indicate the keyword search and the meaning search. In this example, it is shown that the part surrounded by [] is searched for the meaning and the other part is searched for the keyword.

Therefore, it is possible to selectively use the keyword search and the meaning search with one search request, and it is possible to cope with a wider search request.

An embodiment of the information retrieval apparatus according to claim 9 of the present invention will be described below with reference to FIG. In this example,
An object is to provide an information search device that determines a search result and changes a search request when the search result is not satisfied. The same components as those in FIG. 15 are designated by the same reference numerals and the description thereof will be omitted.

The information retrieving apparatus of this embodiment includes a result determining section 41 for determining a retrieval result, and a retrieval request modifying section 42 for revising a retrieval request when the retrieval result is not satisfactory.

Next, the operation of this embodiment will be described with reference to the flowchart of FIG. The same steps as those in FIG. 18 are designated by the same reference numerals and the description thereof will be omitted.

The result judging unit 41 judges whether or not the search result is satisfied (step S50). Whether or not the search result is satisfied can be determined by, for example, a method of presenting the current search result to the user and determining whether or not the user is OK,
If the number of results is equal to or more than a predetermined number and equal to or less than the predetermined number, it is possible to consider a method of automatically determining OK. When the search result is not satisfied, the search request correction unit 42
Thus, the search request is changed (step S51), and the operations after step S42 are repeated. When it is determined that the number of search results is small, a part or all of the portion designated as the keyword search in the input search request is replaced with the semantic search. If it is determined that the number of search results is large, a part or all of the portion designated as the semantic search in the input search request is replaced with the keyword search. In general, the meaning search is performed even if the keywords do not match, so that more search results can be obtained than the keyword search, and thus such a change is effective. Of course, not only changing from the keyword search to the meaning search and the meaning search to the keyword search, but also changing the search threshold in the meaning search and the threshold of the similarity used as the search result may be included. This is because the number of results obtained in the semantic search can be increased or decreased by changing the search threshold.

FIG. 22 shows an example of conditional statement expansion when a search fails. In the example of FIG. 22, there are few search results under the original search condition, and the expected search result was not obtained, so the case where the expansion is performed is shown. Here, the keyword search is entirely changed to semantic search. As a result, it becomes possible to automatically correct the search text, and the trouble of changing the search text can be omitted and an efficient search can be performed.

Therefore, it is possible to perform a search that makes the best use of the advantages of both the keyword search and the meaning search.

[0077]

As described above, according to the information retrieving apparatus of claim 1, it is checked whether or not a plurality of retrieval results include the same result, and the same result is included. In this case, since it is configured that only one result of the same content is left and other results are deleted, the search result of the duplicate content is not output, and the user can search multiple search results of the same content. It is possible to eliminate the waste of looking around.

According to the information retrieving apparatus of the second aspect, the retrieval result of the duplicated content is not output, and the waste of the user seeing the retrieval result of the same content a plurality of times can be omitted.

According to the information retrieving apparatus of the third aspect, since information having a semantic cohesion is continuously presented, the user can easily understand the retrieval result.

According to the information retrieving apparatus of the fourth aspect, the user is automatically requested to make an answer when there is no answer to the retrieval request, and the trouble of replying to the user again can be omitted.

According to the information retrieving apparatus of the fifth aspect, when there is no response to the retrieval request, the user is automatically required to provide the response, and the trouble of re-replying to the user can be omitted.

According to the information retrieving apparatus of claim 6, if the user of the receiving terminal does not want to answer a question from another user, if the user question mode is set to'NO ', the question from another person is asked. I don't have to answer. For example, when the user of the receiving terminal is very busy, the answer is "NO" so that the question is not presented, and when the user is not busy, the answer is "YES" to answer the question. Further, when the user rarely uses the receiving terminal, even if the user asks a question, the opportunity for the user to touch the question is limited, so the mode may be set to “NO”. For example, when the receiving terminal is performing reservation management work in a conference room, there is no specific user who directly operates the terminal, but rather a large number of unspecified information inputs and inquiries. Thus, the device alone manages the allocation of conference rooms and the information related thereto. Even in this case,
Since the person who answers the question to the receiving terminal is unknown, it is possible to prevent confusion by setting the mode to'NO 'and prohibiting the question to the user of the receiving terminal.

According to the information retrieving apparatus of the seventh aspect, the keyword retrieving means performs the retrieval by keyword matching, and the meaning retrieving means retrieves according to the meaning and the content. The search and the semantic search can be used together, and it is possible to meet a wider search request.

According to the information retrieving apparatus of claim 8, it is possible to retrieve the meaning.

According to the information retrieving apparatus of the ninth aspect, it is possible to perform retrieval utilizing the advantages of both the keyword retrieval and the semantic retrieval. For example, even if the keyword search fails, the success rate of the search can be improved by performing the semantic search again.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a terminal of the information search device according to claim 1.

FIG. 2a is a flowchart showing an operation of a search request transmitting side of the information search device of claim 1;

FIG. 2b is a flowchart showing the operation of the search request receiving side of the information search device of claim 1;

FIG. 3 is a block diagram showing a configuration of a terminal of the information search device according to claim 3;

FIG. 4 is a block diagram showing a configuration of a result classification unit.

FIG. 5 is a flowchart showing the operation of the embodiment.

FIG. 6 is a diagram showing a display example of a display unit of a terminal of the information search device according to the present invention.

FIG. 7 is a diagram showing a display example of a display unit of a terminal.

FIG. 8 is a block diagram showing a configuration of a terminal of the information search device according to claim 4;

FIG. 9 is a flowchart showing the operation of the present invention.

FIG. 10 is a diagram showing terminals connected to a network.

FIG. 11 is an example of a display screen of a display unit when a search request is made.

FIG. 12 is an example of a display screen when there are no search items.

FIG. 13 is a block diagram showing a configuration of a terminal of the information search device according to claim 6;

FIG. 14 is a flowchart showing the operation of the present invention.

FIG. 15 is a block diagram showing a configuration of an information search device according to claim 7;

FIG. 16 is a block diagram showing a configuration of a meaning search unit of the information search device of claim 7;

FIG. 17 is a block diagram showing a configuration of a feature quantity dictionary of the information search device according to the seventh aspect.

FIG. 18 is a flowchart showing the operation of the present invention.

FIG. 19 is a diagram showing an example of inputting search conditions.

FIG. 20 is a block diagram showing a configuration of a feature quantity dictionary of the information search device according to claim 9;

FIG. 21 is a flowchart showing the operation of the present invention.

FIG. 22 is a diagram showing an example of conditional sentence expansion when a search fails.

[Explanation of symbols]

 1,33 Input part 2 Network part 3,20,36 Data storage part 4,21 Search part 5,22,39 Search result buffer 6 Duplicate deletion part 7,25,40 Display part 16 Search request input part 17 Search device specification part 18 Search request transmission unit 19 Search request reception unit 23 Search response reception unit 24 Search response transmission unit 26 Search result determination unit 27 Search request display unit 28 User response input unit 29 User response transmission unit 34 Search request interpretation unit 35 Search range setting unit 37 Keyword search section 38 Semantic search section

Claims (9)

[Claims] 1. An input unit for inputting search data, a data storage unit for holding data to be searched, a search unit for performing a search in response to a search request, and a buffer unit for temporarily holding a searched search result. Display means for displaying the searched data and the search results, the search request, and network means for communicating the search results via a network, and a plurality of the search results include search results having the same content. It is provided with a deletion means for checking whether or not the search result is included and for deleting the other search result while leaving only one search result of the search results having the same content when the same is included. Information retrieval device. 2. The information retrieving apparatus according to claim 1, wherein the deleting unit includes a determining unit that determines that the data are the same based on an ID number given to the data when the data is created. 3. An input means for inputting search data, a data storage means for holding searched data, a search means for searching for a search request, and a buffer means for temporarily holding the searched search results. Display means for displaying the searched data and the search results, network means for communicating the search request and the search results via a network, and classification means for classifying a plurality of the searched search results. An information retrieval device comprising: 4. An input means for inputting search data, a designating means for designating another information search device on the network to be searched, and a first request for sending a search request to the designated other information search device. Search means, a first receiving means for receiving the search request via the network, a data storage means for storing search data, and a search for referring to the data storage means based on the search request Means, buffer means for holding the search result, second transmitting means for transmitting the search result through the network, second receiving means for receiving the search result through the network, and displaying the search result And a query means for presenting the search request to the user and receiving a response from the user through the network when there is no search result. An information retrieval device characterized by being obtained. 5. The inquiry means is provided with result determination means for determining whether or not there is the search result, request display means for presenting the search request to the user when the search result is not provided, and The information search device according to claim 4, further comprising: an answer input means for inputting a user's answer to the search request; and an answer sending means for sending the entered answer. 6. The information retrieval device further comprises a mode setting means for setting a user question mode as to whether to ask a question to the user, a mode buffer for storing the user question mode, and the user question mode. The information search device according to claim 4, further comprising: a mode determining unit for determining. 7. An input means for inputting a search request, an interpreting means for interpreting the search request, a data storage means for storing data to be searched, a keyword search means for performing a search by keyword matching, and, depending on the meaning. An information retrieval apparatus comprising: a semantic retrieval means for conducting a retrieval, a retrieval result buffer means for temporarily retaining retrieval results, and a display means for displaying the retrieval results. 8. The meaning retrieving means analyzes a stored document to extract an independent word, and a feature quantity dictionary for a characteristic amount corresponding to the independent word obtained by the document analyzing means. Using a word vector generation means for reading out from the word vector to generate a word vector, a word vector storage means for storing the generated feature amount of the independent word, and one or more word vectors stored in the word vector storage means Sentence vector storage means for obtaining and storing a sentence vector representing the characteristics of the sentence, and the similarity between the feature quantity of the word included in the search request and the feature quantity of the word vector stored in the word vector storage means. The information retrieving apparatus according to claim 7, further comprising a desired similarity extraction unit. 9. The information search apparatus according to claim 7, further comprising: result determination means for determining the search result, and correction means for correcting the search request when the search result is not satisfied. Search device.