JP2005250674A - Information retrieval system and program for information retrieval - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information retrieval system capable of automating threshold settings for narrowing down conception search results. <P>SOLUTION: A full-text search part 4 searches for documents containing a character string that matches a search character string inputted from a search request input part 1. A concept search part 5 searches for documents containing any character strings semantically related to the search character string inputted from the search request input part 1. A control part 9 sees if there is a match between the result of the full-text search by the full-text search part 4 and the result of the concept search by the concept search part 5, determines a threshold based on whether there is the match, selects a concept search result based on the threshold, and outputs the result along with the result of the full-text search. When the threshold has been reset by a reset directing part 8, the result of the concept search is reselected according to the threshold changed. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an information search apparatus and an information search program capable of quickly searching important documents required for business from a large amount of stored documents and utilizing them according to business purposes. Is.

  As computerization of documents progresses, the importance of search methods for acquiring necessary information in a short time is increasing. Among such search methods, a search apparatus including a plurality of search means including a concept search means capable of searching for documents having similar contents even if the expression is different from the input is disclosed in Patent Document 1, for example. was there.

  In the information processing apparatus described in this document, when a search request (query) from a user is obtained, the input text is decomposed into words, and then a full text search, a concept search, or a phrase search is selected by selecting a search method. Search was performed and this was displayed. With such a configuration, when there are two or more search modes, one or more search modes to be applied are determined according to the number of characters included in the query characters and the language analysis result, and the search mode is determined according to the query analysis result. The weighting of the results was determined.

JP 2003-085201 A

However, the conventional information retrieval apparatus has the following problems.
In other words, since the search result depends on the document to be searched, in the conventional method of determining the search mode and weighting based on the analysis result of the query input before obtaining the search result, the document including a character string that does not match the input Since the threshold value of the concept search for obtaining the search result cannot be automatically set, the increase or decrease in the search result is not clear to the user.

  The present invention has been made to solve the above-described problems. Conventionally, it has been unclear to the user how to set the threshold value of the search condition. An object of the present invention is to obtain an information search apparatus and an information search program capable of automating the setting of a threshold for cutting.

  An information search apparatus according to the present invention includes a full-text search unit that searches for a document that matches a character string input by a search request input unit, and a concept search that searches for a document that includes a character string that is semantically related to the input character string A full-text search result and a concept search result are collated, and a threshold for determining which concept search result is selected from the concept search results is determined based on the collation results. It is a thing.

  The information search apparatus according to the present invention compares a full-text search result and a concept search result, and sets a threshold for determining which concept search result is selected from the concept search results based on the comparison results. Since the determination is made, it is possible to automate the setting of a threshold for cutting off the concept search result.

An embodiment of the present invention will be described below.
Embodiment 1 FIG.
1 is a block diagram showing an information retrieval apparatus according to Embodiment 1 of the present invention.
In the figure, an information search apparatus includes a search request input unit 1, a display unit 2, a document registration unit 3, a full text search unit 4, a concept search unit 5, an expansion instruction unit 6, a narrowing instruction unit 7, a reset instruction unit 8, and a control. It consists of part 9.

  The search request input unit 1 is a means for a user to make a search request and to input a character string for which the search is requested, and to pass these data to the control unit 9, for example, an input means such as a keyboard or a pointing device; The display unit 2 has a function of performing display control for inputting a character string and making a search request. Alternatively, the search request input unit 1 may be configured with an interface or the like for inputting a character string to be searched from outside the apparatus. The display unit 2 includes a display device such as a CRT (Cathode Ray Tube) or a liquid crystal display, and is a display unit that displays display control data from the control unit 9. The document registration unit 3 is a registration unit formed on a storage unit such as a hard disk, which has a document to be searched and information such as an index necessary for the search by analyzing the document.

  The full-text search unit 4 is a functional unit that sends information on a document that matches the character string of the search request to the control unit 9. The concept search unit 5 is a functional unit that sends information of documents scored in the order similar to the content of the search request to the control unit 9. Details of the search processing in the full text search unit 4 and the concept search unit 5 will be described in detail in the operation section.

  The expansion instruction unit 6 is an instruction unit for designating whether or not to obtain a search result in a wider range than the full-text search result using the concept search result. Specifically, the extension instruction unit 6 is displayed on the display unit 2. A concept expansion instruction is performed by the user selecting an expansion button. The refinement instruction unit 7 is an instruction unit for designating whether or not to obtain a search result for narrowing down the full-text search result using the concept search result. Specifically, a refinement button displayed on the display unit 2 The user selects the concept and instructs to narrow down the concept.

  The reset instruction unit 8 is an instruction unit for designating whether or not to reset the threshold value after the threshold value is determined based on the collation result. Specifically, the reset instruction unit 8 is displayed on the display unit 2. The user resets the threshold value by changing the slide bar (which will be described later).

  The control unit 9 is a functional unit that manages and manages the information search apparatus, collates the full-text search result of the full-text search unit 4 with the concept search result of the concept search unit 5, and based on these collation results, A threshold for determining which concept search result to select from among the concept search results is determined, and when a reset instruction is given by the reset instruction unit 8, the reset threshold is newly set. A concept search result is selected as a threshold value.

  Further, the information search device described above is realized using a computer. The full-text search unit 4, the concept search unit 5, and the control unit 9 are software corresponding to each function, a CPU, a memory, and the like for executing the software. It consists of hardware.

Next, the operation of the first embodiment will be described.
FIG. 2 is an explanatory diagram illustrating a state of the display unit 2 when the information search apparatus is activated.

  In the figure, a search character string input unit 201 is configured on the display unit 2 corresponding to the search request input unit 1 in FIG. 1, and displays a character string for which a user makes a search request. The search button 202 has a configuration corresponding to the search request input unit 1 in FIG. 1 and is a button for a user to issue a search instruction. The search button 202 is pressed on the display unit 2 (selected with a mouse). The information search is started by performing an operation such as clicking. The expansion button 203 has a configuration on the display unit 2 corresponding to the expansion instruction unit 6 in FIG. 1, and the narrowing button 204 has a configuration on the display unit 2 corresponding to the narrowing instruction unit 7 in FIG.

FIG. 3 is an operation flowchart when the concept expansion instruction according to the first embodiment is performed.
First, as shown in FIG. 2, it is assumed that a search character string such as “connection method with servo” is input to the search character string input unit 201 by the user and the search button 202 is pressed.

  As a result, the control unit 9 issues a search instruction to the full-text search unit 4. As a full-text search step, the full-text search unit 4 searches the text information input from the search request input unit 1 based on the index information from the document registration unit 3 for document information that matches the search request character string. The data is transferred to the control unit 9 (step ST101). The document information is information such as a document number uniquely set for the document. Here, the full-text search is a search for a document including all search character strings. However, since this search method is known, a detailed description thereof is omitted here. In step ST101, if the upper limit of n items for obtaining the search result is specified as a part of the search target, n full-text search results are obtained.

  Next, in step ST102, the control unit 9 determines whether or not the concept extension is designated by the extension instruction unit 6, that is, whether or not the extension button 203 is pressed, and if the concept extension is not designated. Then, the full text search result is displayed on the display unit 2 as the search result as it is (step ST103).

  If the concept extension is designated in step ST102, the concept search unit 5 passes the document information scored in the order similar to the content of the search request to the control unit 9 as a concept search step (step ST102). Step ST104). At this time, if the maximum number m of concept search results to be acquired is specified as a part of the search request, only m search results with higher scores are acquired.

  For the concept search, for example, a method disclosed in literature ("Information search system InfoMAP based on word associations, Takayama et al., Informatics Fundamentals 53-1, 1999-3, Information Processing Society of Japan") is used. In this document, text contained in a registered document is analyzed and divided into words, and an index is generated by algebraic operation called singular value decomposition from the frequency of words that appear (co-occur) at the same time. Is to do.

  Next, in step ST105, the control unit 9 collates the n full-text search results obtained in step ST101 with the m concept search results obtained in step ST104 (search result collation step). In this search result collation step, information such as a document number for uniquely identifying the search result document is used.

  Next, as a threshold determination step, a score s having a lowest concept search score among n full-text search results is obtained as a threshold value determination step from documents matched as a result of collation in step ST105 ( Step ST106). At this time, for the calculation of the score s, a value obtained by performing an operation at the lowest score may be used. For example, a predetermined value is added to the lowest score. This is effective when the minimum score is used depending on the document to be searched, resulting in a large amount of documents obtained as a search result. For example, a value obtained by adding 0.1 to the minimum score is used. The calculation is used as a score.

  Next, the control unit 9 displays on the display unit 2 the result of the concept search having a score higher than the score s from the results of the concept search obtained in step ST104, using the score s obtained in step ST105 as a threshold value. (Step ST107).

  As a result, when “concept expansion” is instructed, the number of search results always increases compared to the full-text search results, and it is possible to obtain search results ranked according to the contents of the document. It becomes.

  In step ST108, the slide bar 205 for resetting the threshold is displayed.

FIG. 4 is an explanatory diagram of a display screen when the threshold is reset.
In the figure, a slide bar 205 is configured on the display unit 2 corresponding to the reset instruction unit 8 in FIG. 4 and is a graphical user interface that visually displays a range of threshold resetting. The display position of the slide bar 205 is a value corresponding to the score s automatically obtained in step ST106 in FIG.

  When the user determines that there are too few search results when the automatically obtained score s is set as a threshold value, the user moves the slide bar 205 in the direction in which “expand” is displayed, and again the search button 202. Is pressed (step ST109). Thereby, the control part 9 changes the value of a threshold value (step ST110), returns to step ST107, and obtains a concept search result with the changed threshold value. Such an operation is repeated as long as the user resets the threshold by moving the slide bar 205.

  As described above, according to the first embodiment, a search request input unit that inputs a character string that requests a search, and a full-text search unit that searches for a document that includes a character string that matches the character string input by the search request input unit And a concept search unit that searches for a document that includes a character string that is semantically related to the character string input in the search request input unit, a full-text search result, and a concept search result are collated, and based on these collation results And a control unit that determines a threshold value for determining which concept search result is to be selected from the concept search results, so that a threshold value for selecting the concept search result can be automatically set. Thus, the user can clearly know the increase or decrease in the search results.

  In addition, according to the first embodiment, the information search program for realizing the operation control of the information search apparatus that performs the full-text search and the concept search of the document on the basis of the character string requested for the search is matched with the character string. Full-text search step for searching for documents containing character strings to be searched, conceptual search step for searching for documents containing character strings semantically related to character strings, full-text search results of full-text search steps, and conceptual search of concept search steps A search result matching step for matching results and a threshold determination step for determining a threshold for determining which concept search results to select from the concept search results based on the results of the search result matching step are realized. It is possible to automatically set a threshold for selecting a concept search result on a computer. It is possible to know clearly the increase or decrease of the search results for the user.

  Further, according to the first embodiment, since the control unit that outputs all of the full-text search results, the concept search result having a value larger than the threshold value, and the display unit that displays the output result of the control unit, the user is provided. Therefore, the search result can be easily confirmed on the display unit. In addition, it is possible to reduce the number of search results to only those whose document content is conceptually similar to the search request and output the search results.

  Also, according to the first embodiment, when there is an extension instruction in the concept extension instruction unit that specifies whether or not to obtain a search result in a wider range than the full-text search result using the concept search result, and the concept extension instruction unit In addition, since the control unit for determining the threshold value is provided, the user can freely set whether to obtain a wide range of search results.

  Further, according to the first embodiment, after the threshold value is determined based on the collation result, a reset instruction unit for specifying whether or not to reset the threshold value, and a reset instruction are issued by the reset instruction unit. In this case, the control unit selects a concept search result using the reset threshold as a new threshold, and the user can arbitrarily change the automatically set threshold. It is possible to easily obtain street search results and obtain search results in a form that is easy for the user to understand.

  In addition, according to the first embodiment, since the search window (search character string input unit 201) that combines the two types of search function calls is combined into one, the user has two types of search functions as in the past. There is no need to consciously use it properly, and the search operation is easy for the user.

  In addition, according to the first embodiment, since the extension operation is instructed by a simple operation for the user such as pressing the search button 202 or the extension button 203, the extension instruction is given in a form that is easy for the user to understand. be able to.

Embodiment 2. FIG.
In the second embodiment, search results are narrowed down in the information search apparatus described in the first embodiment.

  The configuration on the drawing is the same as that of the first embodiment shown in FIG.

Next, the operation of the second embodiment will be described.
FIG. 5 is an operation flowchart when the concept narrowing-down instruction according to the second embodiment is performed.
First, the control unit 9 issues a search instruction to the full-text search unit 4 as in the concept expansion operation. The full text search unit 4 receives the text information input from the search request input unit 1 based on the index information from the document registration unit 3 and receives information on a document that matches the character string of the search request from the control unit 9. (Step ST201).

  Next, in step ST202, the control unit 9 determines whether or not a concept narrowing instruction is given by the narrowing down instruction unit 7, that is, whether or not the narrowing down button 204 is pressed, and if the concept narrowing down is not designated. Then, the full text search result is displayed on the display unit 2 as the search result as it is (step ST203).

  In step ST202, the operation from step ST204 to step ST205 in the case where the concept extension is designated is the same as the operation from step ST104 to step ST105 in the first embodiment, and thus the description thereof is omitted here. To do. Next, in step ST206, as a concept narrowing step, among n documents of the full-text search result, a document having a positive concept search score for the document (having a value larger than the reference value with a reference value of 0) Document) as a result. Here, it is assumed that the score of the concept search takes a value in a range from 1 that completely matches to −1 that is completely different depending on the similarity to the search request.

  In the present embodiment, an intermediate value of the score is set to 0 as a reference value, and a score having a positive value is obtained. However, when the score of the concept search is in another range, a predetermined value is used. It is also possible to employ a configuration in which a document having a degree of similarity equal to or higher than the reference value is obtained as a result based on the value based on the reference.

  Next, in step ST207, the control unit 9 causes the display unit 2 to display the result of the full text search in which the concept search score obtained in step ST206 is positive in the order of the concept search score.

  As a result of such operations, when “concept narrowing” is instructed, the number of search results is always reduced compared to the full-text search results, and the ranking according to the content of the document is made based on the concept search score. Search results can be obtained.

  Even in the case of conceptual narrowing, a slide bar for resetting is displayed together with the display of the search result (step ST208). This slide bar has the same configuration as the slide bar 205 described in FIG. 4 in the first embodiment except that the threshold value is the reference value. If the user determines that there are too many search results by the automatically obtained narrowing down, the user moves the slide bar 205 in the “squeeze” direction and presses the search button 202 again (step ST209). Thereby, the control part 9 changes a reference value (step ST210), and returns to step ST207. By such an operation, it is possible to narrow down search results intended by the user.

  As described above, according to the second embodiment, when there is a refinement instruction unit that specifies whether to obtain a search result that narrows down the full-text search result using the concept search result, and a refinement instruction unit, Since the search result document of the full-text search result is provided with a control unit that obtains a document with a matching value of the concept search result larger than the reference value as a result, the user can easily narrow down the search result. In addition, even if the search result based on the automatically set reference value is different from the intended result, it is possible to obtain a re-search result obtained by reducing the search result as intended.

  Further, according to the second embodiment, upon receiving a narrowing-down instruction for obtaining a search result for narrowing down the full-text search result using the concept search result, the degree of coincidence of the concept search result in the search result document of the full-text search result A program for information retrieval that realizes a concept narrowing step for obtaining documents having a value larger than the reference value as a narrowing search result is provided, so that the user can easily obtain a result of narrowing the search result. A computer capable of realizing a re-search result with a reduced search result as intended even when the search result based on the automatically set reference value is different from the result intended by the user. Can be realized.

Embodiment 3 FIG.
In the third embodiment, the difference between the full text search result and the concept search result is displayed.

FIG. 6 is a configuration diagram of the information search apparatus according to the third embodiment.
The apparatus shown in FIG. 1 includes a search request input unit 1, a display unit 2, a document registration unit 3, a full-text search unit 4, a concept search unit 5, an expansion instruction unit 6, a refinement instruction unit 7, a reset instruction unit 8, a control unit 9, It consists of a difference display instruction unit 10. Here, since the search request input unit 1 to the reset instruction unit 8 are the same as those in the first and second embodiments, the corresponding parts are denoted by the same reference numerals and the description thereof is omitted. Moreover, the difference display instruction | indication part 10 is a function part for instruct | indicating the display of the search result only of the difference of a full-text search result and a concept search result. The basic function of the control unit 9a is the same as that of the control unit 9 in the first embodiment. However, as a function in the second embodiment, when the difference display instruction unit 10 has a difference display instruction, The display unit 2 has a function of displaying only the concept search result of the full text search result and the concept search result.

Next, the operation of the third embodiment will be described.
Since the search result expansion operation and the narrowing-down operation are the same as those in the first and second embodiments, description thereof will be omitted, and a differential display operation when the expansion button 203 is pressed will be described as an example.

FIG. 7 is an explanatory diagram showing a search request character string input screen and search results.
In this example, “servo connection method” is input as a search character string, and the expansion button 203 is pressed and the search button 202 is pressed to display a list of search results.

  Further, in the figure, the difference display button 206 corresponds to the configuration on the display unit 2 of the difference display instruction unit 10 in FIG. 6, and the difference display is performed by pressing the difference display button 206. It has become.

  In the search result shown in FIG. 7, the title of the search result “connection method with the servo motor”, the score of the concept search “82%”, and the summary of the document “in this chapter, with the device…” Each sentence is displayed in order of score. At this time, the input character string itself, for example, a search result including “servo motor” or the like does not include a character string that completely matches the input character string, but it is the same as the input like “small motor”. The search result of the document including the word representing the content is displayed in a mixed form.

  When the difference display button 206 and the search button 202 are pressed in such a state, the control unit 9 searches the search result of the full text search result and the concept search result as the difference result extraction step (see step ST107 in FIG. 3). Among them, only the concept search result from which the full-text search result is removed is displayed on the display unit 2.

FIG. 8 is an explanatory diagram of a search result screen when the difference display is performed.
As shown in the figure, six differences are displayed as only the concept search results obtained by removing the full-text search results from the 27 search results shown in FIG. That is, in this case, only the search result spread by the expansion operation can be displayed, so that the user can easily confirm the search result.

  As described above, according to the third embodiment, when there is a difference display instruction in the difference display instruction unit for instructing to display only the difference between the full text search result and the concept search result, and the difference display instruction unit, Of the full-text search result and the concept search result, the control unit that instructs the display unit to display only the concept search result is provided, so that the user can easily check the search result expanded by the expansion operation.

  Further, according to the third embodiment, the difference display instruction for instructing to display the search result of only the difference between the full-text search result and the concept search result is received, and the concept is selected from the full-text search result and the concept search result. Since the information search program for realizing the difference result extraction step for displaying only the search results on the display unit is provided, a computer that allows the user to easily check the search results expanded by the expansion operation is provided. Can be realized.

  In each of the above embodiments, two types of search results, full text search and concept search, are used, but natural text search and phrase search may be used as an extension of full text search. Here, the natural text search is a search that automatically divides the input of the full text search for each word, and the phrase search is a search that always searches a document in which a plurality of words specified in the full text search are always continuous. Is the method.

It is a block diagram which shows the information search device by Embodiment 1 of this invention. It is explanatory drawing which shows the state of the display part at the time of starting of the information search device by Embodiment 1 of this invention. 6 is an operation flowchart when a concept extension instruction according to the first embodiment is performed. It is explanatory drawing of the display screen at the time of a threshold reset. 10 is an operation flowchart when a concept narrowing-down instruction according to the second embodiment is performed. FIG. 10 is a configuration diagram of an information search device according to a third embodiment. It is explanatory drawing which shows a search request character string input screen and a search result. It is explanatory drawing of the search result screen at the time of performing a difference display.

Explanation of symbols

  1 search request input unit, 2 display unit, 4 full-text search unit, 5 concept search unit, 6 expansion instruction unit, 7 narrowing instruction unit, 8 resetting instruction unit, 9, 9a control unit, 10 differential display instruction unit, 201 search Character string input part 202 Search button 203 Expansion button 204 Narrow down button 205 Slide bar 206 Difference display button

Claims (9)

A search request input part for inputting a search request character string;
A full-text search unit for searching for a document including a character string that matches the character string input in the search request input unit;
A concept search unit for searching for a document including a character string semantically related to the character string input in the search request input unit;
A control unit that collates the full-text search result and the concept search result, and determines a threshold for determining which concept search result is selected from the concept search results based on the collation results; An information retrieval apparatus comprising: A control unit that outputs all of the full-text search results and a concept search result having a value larger than the threshold;
The information search apparatus according to claim 1, further comprising: a display unit that displays an output result of the control unit. A concept extension instruction unit that specifies whether to obtain a search result in a wider range than the full-text search result using the concept search result;
The information search apparatus according to claim 1, further comprising a control unit that determines a threshold value when an extension instruction is given by the concept extension instruction unit. After determining the threshold value based on the collation result, a reset instruction unit for specifying whether or not to reset the threshold value;
The information according to claim 1, further comprising: a control unit that selects a concept search result using the reset threshold as a new threshold when the reset instruction is issued by the reset instruction unit. Search device. A refinement instruction unit that specifies whether to obtain a search result that narrows down the full-text search result using the concept search result;
A control unit for obtaining, as a narrowing search result, a document having a matching degree of a concept search result larger than a reference value in a search result document of a full text search result when a narrowing instruction is given by the narrowing down instruction unit. The information search apparatus according to claim 1, wherein A difference display instruction unit for instructing to display a search result of only the difference between the full-text search result and the concept search result;
The difference display instruction unit includes a control unit that instructs the display unit to display only the concept search result among the full-text search result and the concept search result when there is a difference display instruction. The information search device according to claim 1. A program for information retrieval that realizes operation control of an information retrieval device that performs full-text retrieval and concept retrieval of a document based on a character string requested for retrieval by a computer,
A full-text search step for searching for a document including a character string matching the character string;
A concept search step of searching for a document including a character string semantically related to the character string;
A search result matching step for matching the full-text search result of the full-text search step with the concept search result of the concept search step;
A program for information search that realizes a threshold value determining step for determining a threshold value for determining which concept search result is selected from the concept search results based on the result of the search result matching step.   In response to a narrowing-down instruction for obtaining a search result for narrowing down the full-text search result using the concept search result, a document having a matching score greater than a reference value in the search result document of the full-text search result is selected. The information search program according to claim 7, wherein the concept narrowing step to be obtained as a narrow search result is realized. In response to a difference display instruction for instructing to display a search result of only the difference between the full text search result and the concept search result, a difference in which only the concept search result is displayed on the display unit among the full text search result and the concept search result. 8. The information retrieval program according to claim 7, wherein the result extraction step is realized.

Images