Search effectiveness visualization system, search effectiveness visualization method, and carrier device

Abstract

The search system (10) includes: a search term acquisition unit (111) configured to acquire a search term; a full-text search unit (112) configured to perform a search operation based on the search term; and a visualization unit (114) configured to display a correspondence between the search term and a result obtained by performing the search operation.

Description

Technical Field

[0001] This invention relates to a search effectiveness visualization system, a search effectiveness visualization method, and a carrier device. Background Technology

[0002] Previously, searches were performed based on one or more search terms extracted from one or more natural sentences entered by the user, and the search results were ranked.

[0003] For example, International Application Publication No. WO 2009 / 048130 (PTL1) discloses a technique for calculating the importance of a document as a matching index for search criteria, and then displaying that document. Specifically, as disclosed in paragraph

[0145] and Figure 32 of International Application Publication No. 2009 / 048130, documents drawn in the upper left and lower right of a two-dimensional plane are displayed as documents that match each of the search criteria.

[0004] Citation List

[0005] Patent documents

[0006] [Patent Document 1] International Application Publication No. WO 2009 / 048130 Summary of the Invention

[0007] Technical issues

[0008] However, the documents shown in International Application Publication No. 2009 / 048130 only indicate which document is closer to which search criteria. As a result, users may not understand the validity of the results obtained by performing a search operation.

[0009] The present invention discloses embodiments designed in view of the above-mentioned problems, which are capable of improving the visualization of the effectiveness of the results obtained by performing search operations.

[0010] Solution to the problem

[0011] One aspect of the present invention relates to a search system, comprising: a search term acquisition unit for acquiring search terms; a full-text search unit for performing a search operation based on the search terms; and a visualization unit for displaying the correspondence between the search terms and the results of the search operation.

[0012] Effects of the present invention

[0013] Embodiments of the present invention can improve the visualization of the validity of results obtained by performing search operations. Attached Figure Description

[0014] The accompanying drawings are intended to illustrate exemplary embodiments of the invention and should not be construed as limiting its scope. Unless explicitly stated otherwise, the drawings should not be considered to be drawn to scale. Furthermore, in all the drawings, the same or similar reference numerals denote the same or similar parts.

[0015] Figure 1 The diagram shown is a schematic configuration of a search system including an embodiment of the present invention.

[0016] Figure 2 The diagram shown is an example of the hardware block diagram of a search system according to an embodiment of the present invention.

[0017] Figure 3 The diagram shown is a functional block diagram of the search system according to Embodiment 1 of the present invention.

[0018] Figure 4 The diagram shown is a functional block diagram of the search system according to Embodiment 2 of the present invention.

[0019] Figure 5 The diagram shown is a functional block diagram of the full-text search unit according to Embodiment 2 of the present invention.

[0020] Figure 6 The diagram shown is a functional block diagram of the extended word acquisition unit according to Embodiment 2 of the present invention.

[0021] Figure 7 The diagram shown is an example of the functional block diagram of the visualization unit involved in Embodiment 2 of the present invention.

[0022] Figure 8 The image shown is an example of the data composition information of the search object involved in Embodiments 1 and 2 of the present invention.

[0023] Figure 9 The image shows an example of the data composition information of the search sentence and search results involved in Embodiment 2 of the present invention.

[0024] Figure 10 The image shown is an example of the data composition information of search term influence involved in Embodiment 2 of the present invention.

[0025] Figure 11 The diagram shown is a timing diagram of the search process involved in Embodiment 1 of the present invention.

[0026] Figure 12 The diagram shown is a timing diagram of the search process involved in Embodiment 2 of the present invention.

[0027] Figure 13 The diagram shows a flowchart of the search process involved in Embodiment 1 of the present invention.

[0028] Figure 14 The diagram shows a flowchart of the search process involved in Embodiment 2 of the present invention.

[0029] Figure 15 The image shown is an example of an input screen according to Embodiments 1 and 2 of the present invention.

[0030] Figure 16 The image shown is an example of a search results list according to Embodiment 1 of the present invention.

[0031] Figure 17 The image shown is an example of a search results list according to Embodiment 2 of the present invention.

[0032] Figure 18 The image shown is a visual example of the effectiveness of search results according to Embodiment 2 of the present invention.

[0033] Figure 19 The image shown is a visual example of the correspondence between search terms and search results involved in Embodiment 2 of the present invention.

[0034] Figure 20 The image shown is another example of a visualization of the validity of search results according to Embodiment 2 of the present invention.

[0035] Figure 21 The image shown is an example of a visualization for comparing the effectiveness of search results according to Embodiment 2 of the present invention. Detailed Implementation

[0036] The terms used herein are for the purpose of describing particular embodiments only and are not intended to limit the invention. As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. Specific terms have been used for clarity in describing the embodiments illustrated in the accompanying drawings. However, the disclosure of this specification is not intended to be limited to the selected specific terms, and it should be understood that each specific element includes all technical equivalents that have similar functionality, operate in a similar manner, and achieve similar results. The following description, with reference to the accompanying drawings, describes the configuration for implementing the invention.

[0037] Hereinafter, more than one embodiment will be described with reference to the accompanying drawings. In the drawings and in this invention, the same or similar parts are indicated by the same reference numerals, and repeated descriptions may be omitted.

[0038] (System Composition)

[0039] Figure 1 The diagram shown is a schematic configuration of a search system 10 according to an embodiment of the present invention. Figure 1As shown, the search system 10 is communicatively connected to the input device 20 and the output device 30 via the network 40. The devices and system will be described below.

[0040] Search system 10 is a system that extracts search terms from sentences such as natural sentences (hereinafter referred to as search sentences) input by a user to perform a search operation. In this specification, a search term can be one or more search terms, and a search sentence can be one or more search sentences. Search system 10 can visualize the validity of the results obtained by performing a search operation. Specifically, search system 10 can display the validity of the results obtained by performing a search operation on output device 30. See below for further details. Figure 3 and Figure 4 Let's describe the search system 10 in detail.

[0041] Input device 20 is a device used by the user to input search terms. For example, input device 20 may be a personal computer, tablet computer, smartphone, etc.

[0042] Output device 30 is a device used to display search results obtained by performing a search operation using search system 10. For example, output device 30 may be a personal computer, tablet computer, smartphone, etc.

[0043] Although input device 20 and output device 30 are Figure 1 The input device 20 and the output device 30 are described as different devices, but the input device 20 and the output device 30 can be performed as a single device.

[0044] The apparatus and devices described in the embodiments represent only one example of multiple computing environments in which embodiments of the present invention can be implemented. In one embodiment, the search system 10 may include multiple computing devices as a server cluster. The multiple computing devices are configured to communicate with each other via any type of communication link (including a network or shared memory) and perform the processes disclosed in this specification.

[0045] (Hardware Components)

[0046] Figure 2 The diagram shown is an example of the hardware block diagram of the search system 10.

[0047] like Figure 2As shown, the search system 10 composed of a computer includes, for example, a central processing unit (CPU) 1001, a read-only memory (ROM) 1002, a random access memory (RAM) 1003, a hard disk (HD) 1004, a hard disk drive (HDD) controller 1005, a display 1006, an external device connection interface (I / F) 1007, a network interface (I / F) 1008, a data bus 1009, a keyboard 1010, a pointing device 1011, a rewritable digital multifunction disc (DVD-RW) drive 1013, and a media interface (I / F) 1015.

[0048] CPU 1001 controls the operation of the search system 10. ROM 1002 stores programs that drive CPU 1001, such as the Initial Program Loader (IPL). RAM 1003 is used as the working area of ​​CPU 1001. HD 1004 stores various data such as programs. HDD controller 1005 controls the reading and writing of various data in HD 1004 under the control of CPU 1001.

[0049] The display 1006 displays various information such as cursor, menus, windows, characters, or images. The external device connection interface I / F 1007 is an interface for connecting various external devices. Examples of external devices include Universal Serial Bus (USB) memory, printers, etc. The network I / F 1008 is an interface for performing data communication using the communication network 40. The data bus 1009 is used for electrical connections such as… Figure 2 The address bus and data bus of each component of the CPU1001 are shown.

[0050] Keyboard 1010 is an input unit with multiple keys for inputting characters, numbers, various instructions, etc. Pointing device 1011 is an input unit used for selecting and executing various instructions, selecting processing targets, moving the cursor, etc.

[0051] DVD-RW drive 1013 controls the reading and writing of various data for DVD-RW 1012, which is used as an example of a removable recording medium. In addition to or instead of DVD-RW, the computer may include a recordable digital multifunction disc (DVD-R). Media interface I / F 1015 controls the reading and writing (storage) of data for recording medium 1014, such as flash memory.

[0052] (Functional Composition)

[0053] The following is with reference to the appendix. Figure 3 and Figure 4 The functional configuration of the search system 10 as Embodiment 1 and Embodiment 2 will be described respectively.

[0054] (Example 1)

[0055] Figure 3 The diagram shown is a functional block diagram of the search system 10 according to Embodiment 1 of the present invention. Figure 3 As shown, for example, the search system 10 includes a search term acquisition unit 101, a full-text search unit 102, a search result acquisition unit 103, a visualization unit 104, and a search sentence correction unit 105. Furthermore, the search system 10 can execute one or more programs to function as the search term acquisition unit 101, the full-text search unit 102, the search result acquisition unit 103, the visualization unit 104, and the search sentence correction unit 105. Each unit will be described below.

[0056] The search term acquisition unit 101 acquires search terms for performing a search operation. Specifically, the search term acquisition unit 101 receives a natural sentence (i.e., a search sentence) input by the user from the input device 20 (see [link to relevant documentation]). Figure 3 (1)). Then, the search term acquisition unit 101 segments the received search sentence into each word to extract the search term.

[0057] Then, the search term acquisition unit 101 stores the search terms in the memory so that the full-text search unit 102 and the visualization unit 104 can refer to the search terms (see...). Figure 3 (2) and (4)).

[0058] In addition, the search term acquisition unit 101 can acquire or obtain search terms by receiving search terms input by the user from the input device 20.

[0059] Then, the full-text search unit 102 performs a search operation based on the search terms. Specifically, the full-text search unit 102 uses the search terms to generate a query for performing the search operation. Furthermore, the full-text search unit 102 searches for one or more documents (i.e., search targets) based on the query. Additionally, the full-text search unit 102 stores the search results obtained by performing the search operation in a memory so that the search result acquisition unit 103 can refer to the search results.

[0060] Then, the search result acquisition unit 103 acquires or obtains the search results obtained by performing a search operation using the full-text search unit 102 (see [link]). Figure 3 (3) in the middle), and then send the search results to the visualization unit 104 (see (see Figure 3 (4-1) in the middle.

[0061] Then, visualization unit 104 creates a graph that visualizes the correspondence between documents and search terms (e.g., three or more search terms), and sends the graph to output device 30 (see...). Figure 3(5) in the middle.

[0062] The search query correction unit 105 corrects or modifies the query used to perform the search operation (e.g., changes, additions, or deletions of search terms), and then performs the next search operation based on the corrected or modified query.

[0063] (Example 2)

[0064] Figure 4 The diagram shown is a functional block diagram of the search system 10 according to Embodiment 2 of the present invention.

[0065] In the above embodiment 1, the search operation is performed based on the search terms extracted from the search sentence (or search terms entered by the user).

[0066] In Example 2, a search operation is performed based on a search term (hereinafter referred to as the original search term or initial search term) and other search terms derived from the original search term (hereinafter referred to as extended terms). Hereinafter, the combination of the original search term and the extended terms can be referred to as the new search term.

[0067] Expanded terms are terms used to prepare a query to perform a search operation on one or more documents requested by a user. For example, expanded terms are terms extracted from a document that was used to perform a search operation based on the original search terms by applying pseudo-relevance feedback (PRF) (hereinafter referred to as the initial search operation).

[0068] In addition, expanded terms can be various types of words, such as synonyms of the search term, words representing the broader (higher) scope of the search term, words representing the narrower (lower) scope of the search term, and words that co-occur with the search term. For example, if the search term is "multifunction machine," then the synonym is "MFP," the word representing the broader scope is "office machine," the word representing the narrower scope is "laser printer," and the words that co-occur with the search term are "copy," "block," etc.

[0069] like Figure 4 As shown, for example, the search system 10 includes a search term acquisition unit 111, a full-text search unit 112, a search result acquisition unit 113, a visualization unit 114, a search sentence correction unit 115, and an expanded term acquisition unit 116. Alternatively, the search system 10 can function as the search term acquisition unit 111, full-text search unit 112, search result acquisition unit 113, visualization unit 114, search sentence correction unit 115, and expanded term acquisition unit 116 by executing one or more programs. Each unit will be described below.

[0070] The search term acquisition unit 111 acquires search terms for performing a search operation. Specifically, the search term acquisition unit 111 receives a natural sentence (i.e., a search sentence) input by the user from the input device 20 (see [link to relevant documentation]). Figure 4 (1)). Then, the search term acquisition unit 111 segments the received search sentence into each word to extract the search term. Then, the search term acquisition unit 111 stores the search term in the memory so that the full-text search unit 112 and the visualization unit 114 can refer to the search term (see (1)). Figure 4 (2) and (7-1)).

[0071] In addition, the search term acquisition unit 111 can acquire or obtain search terms by receiving search terms input by the user from the input device 20.

[0072] Unit 116, which acquires or obtains extended words, will be referred to later. Figure 6 Let's describe the extended word acquisition unit 116 in detail.

[0073] Full-text search unit 112 performs search operations based on the original search terms (see...). Figure 3 The full-text search unit 112 also performs search operations based on new search terms (i.e., original search terms and expanded terms). Specifically, the full-text search unit 112 uses new search terms (i.e., original search terms and expanded terms) to generate a query for performing the search operation. Furthermore, the full-text search unit 112 searches for one or more documents (search targets) based on the query. Additionally, the full-text search unit 112 stores the search results obtained through performing the search operation in a memory so that the search result acquisition unit 113 can refer to the search results (see...). Figure 4 (3) and (6) in the text). (See below for reference.) Figure 5 Let's describe the full-text search unit 112 in detail.

[0074] Then, the search result acquisition unit 113 acquires or obtains the search results obtained by performing a search operation using the full-text search unit 112 (see [link]). Figure 4 (3) and (6) in the middle), and then send the search results to the visualization unit 114 (see Figure 4 (7-2) in the middle.

[0075] Then, visualization unit 114 creates a graph that visualizes the correspondence between documents and new search terms (e.g., three or more new search terms consisting of the original search term and expanded terms, respectively), and sends the graph to output device 30 (see...). Figure 4 (8) in the middle). (See section (8) for details.) Figure 7 Let's describe the full-text search unit 114 in detail.

[0076] Then, the search sentence correction unit 115 corrects or modifies the query used to perform the search operation (e.g., changes, additions, or deletions of search terms), and then performs the next search operation based on the corrected or modified query.

[0077] Figure 5 The diagram shown is a functional block diagram of the full-text search unit 112 according to Embodiment 2 of the present invention. The full-text search unit 112 includes, for example, a search term receiving unit 201, an extended term receiving unit 202, a query generation unit 203, a search target collection unit 204, an index storage unit 205, a search unit 206, and a search result sending unit 207. Each unit will be described below.

[0078] Search term receiving unit 201 receives search terms from search term acquisition unit 111 (see...) Figure 4 (2) in the middle.

[0079] The extended word receiving unit 202 receives extended words from the extended word acquisition unit 116 (see...) Figure 4 (5) in the middle.

[0080] The query generation unit 203 uses the search terms (i.e., the original search terms) received by the search term receiving unit 201 and the expanded terms received by the expanded term receiving unit 202 to generate a query for performing a search operation. For example, the query generation unit 203 uses search operators to connect or link weighted search terms (i.e., the original search terms) and expanded terms to generate a query for performing a full-text search. Additionally, when performing an initial search operation, the query generation unit 203 uses the search terms (i.e., the original search terms) received by the search term receiving unit 201 to generate a query for performing the initial search operation.

[0081] Search target collection unit 204 collects one or more documents that can become candidates for search results (i.e., search targets) and stores the one or more documents in index storage unit 205.

[0082] Index storage unit 205 stores one or more documents (i.e., search targets). Examples 1 and 2 can be applied to searching for any document stored in index storage unit 205, as well as to searching for information available on the Internet, such as web pages and websites.

[0083] Search unit 206 searches for documents (i.e., search targets) stored in index storage unit 205 based on the query generated by query generation unit 203, and sets or assigns a sort to each document. For example, search unit 206 uses a method (e.g., Okapi BM25) to sort the documents extracted by performing a search operation based on the relevance between the query and each document.

[0084] The search result sending unit 207 sends the results obtained by the search unit 206 in performing the search operation to the search result obtaining unit 113 (see [link]). Figure 4 (3) and (6)). For example, the search result sending unit 207 sends the search results (i.e., the document information extracted by performing the search operation and the sorting of each document) to the search result obtaining unit 113.

[0085] In addition, in the case of Embodiment 1, since the full-text search unit 102 did not receive any extended words, the full-text search unit 102 performs the search operation only based on the search words received from the search word acquisition unit 101.

[0086] Figure 6 The diagram shown is a functional block diagram of the extended word acquisition unit 116 according to Embodiment 2 of the present invention. The extended word acquisition unit 116 includes, for example, an initial result receiving unit 301, a candidate word calculation unit 302, an extended word selection unit 303, and an extended word sending unit 304. Each unit will be described below.

[0087] The initial result receiving unit 301 receives the results obtained by performing an initial search operation (i.e., a search operation using the original search terms) from the search result obtaining unit 113 (see [link to original search results]). Figure 4 (4)). For example, the information of the document received by the initial result receiving unit 301 is extracted by performing an initial search operation by applying pseudo-relevance feedback (PRF).

[0088] The candidate word calculation unit 302 extracts one or more candidates for extended words based on the results of the initial search operation received from the initial result receiving unit 301, and sets or assigns sorting to the candidates for each extended word.

[0089] The extended word selection unit 303 selects a given number of extended words ranked at the highest position from one or more candidates of extended words sorted by the candidate word calculation unit 302 (for example, the number can be specified by the user or a preset number).

[0090] The extended term sending unit 304 sends the extended terms selected by the extended term selection unit 303 to the full-text search unit 112 and the visualization unit 114 (see reference). Figure 4 (5), (7-3)).

[0091] Figure 7 The diagram shown is a functional block diagram of the full-text search unit 114 according to Embodiment 2 of the present invention. The visualization unit 114 includes, for example, a text receiving unit 401, a temporary storage unit 402, an influence calculation unit 403, a visualization graph creation unit 404, and a visualization graph sending unit 405. Each unit will be described below.

[0092] The text receiving unit 401 receives information about extended terms from the extended term acquisition unit 116 and stores this information in the temporary storage unit 402. Additionally, the text receiving unit 401 receives information about the original search term (i.e., the search term used to derive the extended terms) from the search term acquisition unit 111 and stores this information in the temporary storage unit 402. Furthermore, the text receiving unit 401 receives search results from the search result acquisition unit 113 and stores these search results in the temporary storage unit 402. The search results include information about the documents extracted from the search operation performed based on the original search term and the extended terms, as well as the ranking of each document (see [link to relevant documentation]). Figure 4 (7-2)).

[0093] Temporary storage unit 402 stores information about the expanded terms, information about the original search terms, and search results received by text receiving unit 401.

[0094] The influence calculation unit 403 calculates the influence (e.g., occurrence count, vector similarity) of new search terms (i.e., original search terms and expanded terms) for each document extracted by performing a search operation.

[0095] The visualization creation unit 404 uses visual metrics such as distance and size to create a graph representing the degree of influence based on the influence calculated by the influence calculation unit 403.

[0096] The visualization sending unit 405 sends the visualization created by the visualization creation unit 404 to the output device 30 (see reference). Figure 4 (8)

[0097] In the case of Embodiment 1, since the visualization unit 104 did not receive any extended words, the visualization unit 104 creates a graph based on the search terms obtained or acquired from the search term acquisition unit 101 and the search results received from the search result acquisition unit 103.

[0098] (Data composition information)

[0099] Figure 8 The image shows an example of the data composition information of the search object involved in Embodiments 1 and 2 of the present invention. The index storage unit 205 stores information such as... Figure 8 The data for the search target is shown. For example... Figure 8 As shown, index storage unit 205 stores various information, such as document identifiers or identifiers (e.g., document ID), document titles, and document content (e.g., text). Documents may include, for example, meeting minutes, reports, and daily memos accumulated by users. The document content (e.g., text) becomes the search target.

[0100] Figure 9 The image shows an example of the data composition information of the search sentence and search results involved in Embodiment 2 of the present invention. Temporary storage unit 402 stores information such as... Figure 9 The search terms and search results shown are data. Figure 9 This is an example of the data constituting information used in Example 2. For example... Figure 9 As shown, the temporary storage unit 402 stores new search terms (i.e., original search terms and expanded terms), the weight of each new search term, and the search results obtained by using the query generated by the new search terms (i.e., information of documents extracted by performing a search operation using the original search terms and expanded terms, and the ranking of each document).

[0101] In the case of Example 1, since no extended terms are stored, the search terms and search results are stored.

[0102] The following explains the weighting of original search terms and expanded terms. In embodiments of the present invention, a search operation can be performed by setting weights for original search terms and expanded terms. The weights can be user-specific or specified values, or they can be pre-set values. Furthermore, all original search terms can be given the same weight, or each original search term can be given a different weight. Similarly, all expanded terms can be given the same weight, or each expanded term can be given a different weight.

[0103] Figure 10 The image shows an example of the data composition information of search term influence according to Embodiment 2 of the present invention. The influence calculation unit 403 calculates... Figure 10 The influence of the search terms shown. Figure 10 This is an example of the data constituting information used in Example 2.

[0104] like Figure 10 As shown, the influence calculation unit 403 calculates the influence of the new search term (i.e., the original search term and the expanded term) for each document extracted by performing a search operation. For example, the influence can be the number of times the new search term (i.e., the original search term and the expanded term) appears in each document (i.e., the number of occurrences of the new search term, such as the frequency of occurrence of the new search term), and the similarity between each document and the new search term (i.e., the original search term and the expanded term) (i.e., vector similarity).

[0105] In the case of Example 1, since the influence of extended terms is not calculated, only the influence of search terms is calculated.

[0106] (Handling method)

[0107] The timing of the search processing performed by the search system 10 in Embodiment 1 and Embodiment 2 will be described below.

[0108] Figure 11 The diagram shown is a timing diagram of the search process involved in Embodiment 1 of the present invention.

[0109] The user inputs a natural sentence (i.e., the search sentence) into the input device 20 (1).

[0110] Then, the input device 20 requests the search term acquisition unit 101 to acquire the search term (1.1).

[0111] Then, the search term acquisition unit 101 requests the full-text search unit 102 to perform a search operation (1.1.1).

[0112] Then, the full-text search unit 102 requests the search result acquisition unit 103 to acquire the result of the search operation (2).

[0113] Then, the search result acquisition unit 103 requests the visualization unit 104 to create a graph (2.1) that visualizes the correspondence between documents and search terms.

[0114] Then, the search term acquisition unit 101 requests the visualization unit 104 to create a graph (3) that visualizes the correspondence between documents and search terms.

[0115] Then, visualization unit 104 creates graph (4).

[0116] Then, the visualization unit 104 requests the output device 30 to output the diagram (5).

[0117] Then, the user requests the output device 30 to provide the validity of the search operation (6).

[0118] Then, the user instructs the search sentence correction unit 105 to correct or modify the query used for the search operation via the output device 30 to perform the next search operation (7).

[0119] Figure 12 The diagram shown is a timing diagram of the search process involved in Embodiment 2 of the present invention.

[0120] The user inputs a natural sentence (i.e., a search sentence) into the input device 20 (1).

[0121] Then, the input device 20 requests the search term acquisition unit 111 to acquire the search term (1.1).

[0122] Then, the search term acquisition unit 111 requests the full-text search unit 112 to perform the initial search operation (i.e., the search operation based on the original search term) (1.1.1).

[0123] Then, the full-text search unit 112 requests the search result acquisition unit 113 to obtain the result of the initial search operation (1.1.1.1).

[0124] Then, the search result acquisition unit 113 requests the extended word acquisition unit 116 to acquire the extended word (1.1.1.1.1).

[0125] Then, the extended term acquisition unit 116 requests the full-text search unit 112 to perform a re-search operation (i.e., a search operation based on the original search term and the extended term) (1.1.1.1.1.1).

[0126] Then, the full-text search unit 112 requests the search result acquisition unit 113 to obtain the result of the re-search operation (1.1.1.1.1.1.1).

[0127] Then, the search result acquisition unit 113 requests the visualization unit 114 to create a graph (1.1.1.1.1.1.1.1.1) that visualizes the correspondence between documents and new search terms (i.e., original search terms and expanded terms).

[0128] Then, the search term acquisition unit 111 requests the visualization unit 114 to create a graph (2) that visualizes the correspondence between the document and the original search term.

[0129] Then, the extended word acquisition unit 116 requests the visualization unit 114 to create a graph (3) that visualizes the correspondence between the document and the extended words.

[0130] Then, visualization unit 114 creates graph (4).

[0131] Then, the visualization unit 114 requests the output device 30 to output the diagram (5).

[0132] Then, the user requests the output device 30 to provide the validity of the search operation (6).

[0133] Then, the user instructs the search sentence correction unit 115 to correct or modify the query used for the search operation via the output device 30 to perform the next search operation (7).

[0134] The following is for reference Figure 13 The flowcharts for the search processing of the search system 10, which are examples of Example 1 and Example 2, will be described respectively.

[0135] Figure 13 The diagram shows a flowchart of the search process involved in Embodiment 1 of the present invention.

[0136] In step S10, the search term acquisition unit 101 determines whether the search term received from the input device 20 is a new search term. Specifically, the search term acquisition unit 101 determines whether the search term received from the input device 20 has been received previously or in the past.

[0137] If the search term acquisition unit 101 determines that the search term received from the input device 20 is a new search term (step S10 is "Yes"), the timing proceeds to step S11. If the search term acquisition unit 101 determines that the search term received from the input device 20 is not a new search term (step S10 is "No"), the timing proceeds to step S12.

[0138] Alternatively, the system can be configured to extract search terms from all search sentences without performing step S10, which involves determining whether the received search sentence is a new search sentence.

[0139] In step S11, the search term acquisition unit 101 extracts search terms from the search sentences received in step S10.

[0140] In step S12, the full-text search unit 102 performs a search operation based on the search terms extracted in step S11 (or search terms extracted previously or in the past).

[0141] In step S13, the visualization unit 104 determines whether a request is made to display the validity of the search operation. Specifically, the visualization unit 104 determines whether the user has requested the validity of the search operation via the output device 30.

[0142] If the visualization unit 104 determines that it is required to display the validity of the search operation (step S13: Yes), the timing proceeds to step S14. If the visualization unit 104 determines that it is not required to display the validity of the search operation (step S13: No), the timing proceeds to step S17.

[0143] In step S14, the visualization unit 104 calculates the influence of the search terms on each document extracted by performing the search operation, such as the number of occurrences and vector similarity.

[0144] In step S15, the visualization unit 104 creates a graph representing the degree of influence based on the influence calculated in step S14, using visual metrics such as distance and size.

[0145] In step S16, the search sentence correction unit 105 corrects or modifies the query used to perform the search operation (e.g., changes, additions, or deletions of search terms), and then performs the next search operation based on the corrected or modified query.

[0146] In step S17, the search sentence correction unit 105 determines whether the search operation is complete. If the search sentence correction unit 105 determines that the search operation has been completed (step S17: Yes), the timing ends. If the search sentence correction unit 105 determines that the search operation has not been completed (step S17: No), the timing returns to step S10.

[0147] Figure 14 The diagram shows a flowchart of the search process involved in Embodiment 2 of the present invention.

[0148] In step S20, the search term acquisition unit 111 determines whether the search term received from the input device 20 is a new search term. Specifically, the search term acquisition unit 111 determines whether the search term received from the input device 20 has been received previously or in the past.

[0149] If the search term acquisition unit 111 determines that the search term received from the input device 20 is a new search term (step S20 is "Yes"), the timing proceeds to step S21. If the search term acquisition unit 111 determines that the search term received from the input device 20 is not a new search term (step S20 is "No"), the timing proceeds to step S22.

[0150] Alternatively, the system can be configured to extract search terms from all search terms without performing step S20, which involves determining whether the received search term is a new search term.

[0151] In step S21, the search term acquisition unit 111 extracts search terms (i.e., original search terms or initial search terms) from the search sentences received in step S20.

[0152] In step S22, the full-text search unit 112 performs an initial search operation based on the search terms extracted in step S21 (or search terms extracted previously or in the past).

[0153] In step S23, the extended word acquisition unit 116 extracts extended words from the documents extracted by performing the initial search operation in step S22.

[0154] In step S24, the full-text search unit 112 performs a re-search operation based on the search terms extracted in step S21 (or search terms extracted previously or in the past) and the expanded terms extracted in step S23.

[0155] In step S25, the visualization unit 114 determines whether a request is made to display the validity of the search operation. Specifically, the visualization unit 114 determines whether the user has requested the validity of the search operation via the output device 30.

[0156] If the visualization unit 114 determines that it is required to display the validity of the search operation (step S25: Yes), the timing proceeds to step S26. If the visualization unit 104 determines that it is not required to display the validity of the search operation (step S25: No), the timing proceeds to step S29.

[0157] In step S26, the visualization unit 114 calculates the influence of the search terms (i.e., the original search terms) and the influence of the expanded terms for each document extracted by performing the re-search operation, such as the number of occurrences and vector similarity.

[0158] In step S25, visualization unit 114 creates a graph representing the degree of influence based on the influence of the search term (i.e., the original search term) and the influence of the extended term calculated in step S26, using visual metrics such as distance and size.

[0159] In step S28, the search sentence correction unit 115 corrects or modifies the query used to perform the search operation (e.g., changes, additions, or deletions of search terms), and then performs the next search operation based on the corrected or modified query.

[0160] In step S29, the search sentence correction unit 115 determines whether the search operation is complete. If the search sentence correction unit 115 determines that the search operation has been completed (step S29: Yes), the timing ends. If the search sentence correction unit 115 determines that the search operation has not been completed (step S29: No), the timing returns to step S20.

[0161] The following is for reference Figure 15 Here is an example of the screen displayed on the input device 20 and the output device 30.

[0162] Figure 15 The image shown is an example of an input screen according to Embodiments 1 and 2 of the present invention. The user can use the input screen displayed on the input device 20 to input a search query. Pressing the search button ( Figure 15 If you press the PRF parameter setting button, a search operation will be performed. Figure 15 Once the input screen is switched, it will switch to another screen for setting the weights of the original search terms and expanded terms.

[0163] Figure 16 The image shown is an example of a search results list according to Embodiment 1 of the present invention. Users can confirm search results by viewing the search results list displayed on the output device 30. Figure 16 As shown, the search results list displays the search terms extracted from the search query and the search results (i.e., information about the documents extracted by performing the search operation, and the ranking of each document).

[0164] Figure 17The image shown is an example of a search results list displayed according to Embodiment 2 of the present invention. Users can confirm search results by viewing the search results list displayed on the output device 30. Figure 17 As shown, the search results list displays the search terms extracted from the search query (i.e., the original search terms), the expanded terms extracted from the documents extracted by performing the initial search operation, and the search results (i.e., information about each document extracted by performing a search operation based on the original search terms and expanded terms, and the order of each document).

[0165] The following is with reference to the appendix. Figure 16 and Figure 17 This is used to explain the correction of queries used for search operations. For example, the search results list ( Figure 16 and Figure 17 The screen can be configured to display buttons for correcting or modifying the query, such as buttons for changing search terms, adding search terms, and deleting search terms. Pressing any of these buttons causes the search term correction unit 105 or 115 to correct or modify the query used to perform the search operation (e.g., changing, adding, or deleting search terms), and then perform the next search operation based on the corrected or modified query.

[0166] Figure 18 The image shown is an example of a screen displaying the validity of search results according to Embodiment 2 of the present invention. For example, when the validity button for each document is pressed (see...), Figure 16 and Figure 17 ), just like Figure 18 The diagram shown visualizes the correspondence between each document and each new search term (i.e., the original search term and the expanded term). In the case of Example 1, the expanded terms are not displayed.

[0167] like Figure 18 As shown, document sorting, document ID, and title are displayed at the center of concentric circles, and new search terms (i.e., original search terms and expanded terms) included in the query used for the search operation that has extracted relevant documents are displayed at specific positions within the concentric circles. The method for creating the graph by visualization unit 114 (104) will be described in detail below. Visualization units 104 and 114 perform similar processes, with visualization unit 104 performing specific processes of Embodiment 1 and visualization unit 114 performing specific processes of Embodiment 2, although they are not entirely the same.

[0168] (1) Visualization unit 114 (104) creates two or more graphics (e.g., concentric circles) that share a common center. For example, the concentric circles are arranged at equal intervals.

[0169] (2) The visualization unit 114 (104) places the original search term and the expanded term at specific positions within a concentric circle based on the vector similarity between the original search term and the expanded term. Specifically, the higher the vector similarity between the original search term and the expanded term, the closer the visualization unit 114 (104) places the original search term and the expanded term at specific positions closer to the center. In the case of Embodiment 1, the visualization unit 114 (104) only processes the original search term.

[0170] Specifically, visualization unit 114 (104) places original search terms and extended terms with vector similarities within a specified range in each region between one circle and the next. For example, original search terms or extended terms with vector similarities of "0 to 0.05", "0.05 to 0.10", and "0.10 to 0.15" are arranged in this order, starting from the outer region. That is, original search terms or extended terms with vector similarities within the same range are placed in the same region. In addition, visualization unit 114 (104) determines the placement position within each region based on the similarity between the original search terms and extended terms.

[0171] (3) The visualization unit 114 (104) displays the original search term and the expanded term by changing the display style based on the number of times the original search term and the expanded term appear. Specifically, the more times the original search term and the expanded term appear, the larger the character size of the original search term and the expanded term displayed by the visualization unit 114 (104).

[0172] (4) The visualization unit 114 (104) displays the original search term and the expanded term in a way that distinguishes them by changing the display style. For example, the visualization unit 114 (104) sets an underline for the expanded term or uses different colors to display the original search term and the expanded term.

[0173] Figure 19 The image shown is a visual example of the correspondence between search terms and search results according to Embodiment 2 of the present invention. For example, when the search term button or the expanded term button (see...) is pressed... Figure 16 and Figure 17 ), just like Figure 19 The diagram shown visualizes the correspondence between each document and each new search term (i.e., the original search term and the expanded term). In the case of Example 1, the expanded terms are not displayed.

[0174] like Figure 19 As shown, the search term or expanded term is displayed at the center of the concentric circles, and each document extracted based on the query for a search operation that includes the original search term and expanded terms is displayed at a specific position within the concentric circles. The method for creating the graph by the visualization unit 114 (104) is described in detail below.

[0175] (1) Visualization unit 114 (104) creates two or more graphics (e.g., concentric circles) that share a common center. For example, the concentric circles are arranged at equal intervals.

[0176] (2) The visualization unit 114 (104) places each document at a specific position within a concentric circle based on the vector similarity between the original search term and the expanded term. Specifically, the higher the vector similarity between the original search term and the expanded term, the closer the visualization unit 114 (104) places each document at a specific position closer to the center. In the case of Embodiment 1, the visualization unit 114 (104) only processes the original search term.

[0177] Specifically, visualization unit 114 (104) places each document within a specified range of vector similarity in each region between one circle and the next. For example, documents with vector similarities of "0 to 0.05", "0.05 to 0.10", and "0.10 to 0.15" are arranged in this order, starting from the outer region. That is, documents with the same range of vector similarity are placed in the same region. In addition, visualization unit 114 (104) determines the placement position within each region based on the document similarity.

[0178] (3) The visualization unit 114 (104) displays the document by changing the display style based on the occurrence frequency of the original search term and the expanded term. Specifically, the more times the original search term and the expanded term occur, the larger the character size of the document displayed by the visualization unit 114 (104).

[0179] Figure 20 The image shown is another example of a visualization of the effectiveness of search results according to Embodiment 2 of the present invention. In Embodiment 2 of the present invention, an Archimedean spiral can be used instead of... Figure 18 , Figure 19 , Figure 21 The concentric circles shown are used to visualize the correspondence between each document and each new search term (i.e., the original search term and the expanded term). The Archimedes spiral, because it allows for more precise setting of the distance to the center, enables a more easily understood style of displaying the correspondence between documents and new search terms (i.e., the original search term and the expanded term).

[0180] Additionally, when using the Archimedes spiral, the frequency of occurrence of a new search term (i.e., the original search term and expanded terms) can be expressed by setting or changing the size of the characters in the new search term. Furthermore, the Archimedes spiral allows for the differentiation of new search terms (i.e., the original search term and expanded terms).

[0181] Figure 21The image shown is an example of a screen display for comparing the validity of search results according to Embodiment 2 of the present invention. For example, when multiple validity buttons (e.g., two buttons) are pressed (see... Figure 16 , Figure 17 ),like Figure 21 As shown, a screen will appear displaying the validity of the visualized search results. Pressing the validity button for each document will... Figure 21 The diagram shown visualizes the correspondence between each document and each new search term (i.e., the original search term and expanded terms). This structure allows users to compare the correspondence between multiple documents and new search terms (i.e., the original search term and expanded terms).

[0182] As mentioned above, Figures 18 to 21 This shows an example of a display pattern used to show the results obtained by performing a search operation and the search terms used for the search operation.

[0183] As described in Examples 1 and 2 above, the correspondence between each document and each search term can be visualized, as can the correspondence between each document and each new search term (i.e., the original search term and the expanded term). Therefore, users can correct or modify search terms based on the correspondence, thereby improving the accuracy of the search operation.

[0184] Furthermore, each function of the above embodiments can be implemented by one or more processing circuits or circuits. The processing circuit includes a programmable processor, which includes circuitry. The processing circuit also includes devices such as application-specific integrated circuits (ASICs), digital signal processors (DSPs), field-programmable gate arrays (FPGAs), and conventional circuit elements arranged to perform the functions described above. The program for executing the above embodiments can be stored in a carrier device or storage medium.

[0185] The above embodiments are merely examples and are not intended to limit the scope of the invention. In view of the above teachings, various additional modifications and variations are possible. Therefore, it should be understood that the disclosure of this specification can be practiced in ways different from those specifically described herein, within the scope of the appended claims.

[0186] This patent application is based on and claims priority to Japanese Patent Application No. 2019-215838, filed with the Japan Patent Office on November 28, 2019, the entire disclosure of which is incorporated herein by reference.

[0187] List of reference numerals

[0188] 10 Search System

[0189] 20 Input Devices

[0190] 30 Output devices

[0191] 40 Network

[0192] 101 Search Term Acquisition Unit

[0193] 102 Full-Text Search Unit

[0194] 103 Search Result Acquisition Unit

[0195] 104 Visualization Units

[0196] 105 Search Sentence Correction Unit

[0197] 111 Search Term Acquisition Unit

[0198] 112 Full-text Search Unit

[0199] 113 Search Result Acquisition Unit

[0200] 114 Visualization Units

[0201] 115 Search Sentence Correction Unit

[0202] 116 Extended Word Acquisition Unit

[0203] 201 Search Term Receiving Unit

[0204] 202 Extended word receiving unit

[0205] 203 Query Generation Unit

[0206] 204 Search Target Collection Unit

[0207] 205 index storage units

[0208] 206 Search Units

[0209] 207 Search Result Sending Unit

[0210] 301 Initial Result Receiving Unit

[0211] 302 Candidate Word Calculation Unit

[0212] 303 Expanded Word Selection Unit

[0213] 304 Extended Word Sending Unit

Claims

1. A search system (10) comprising: an input device configured to receive a search sentence; a search term acquisition unit (111) configured to acquire one or more search terms from the search sentence; a full-text search unit (112) configured to perform a search operation based on the one or more search terms; and a visualization unit (104, 114) configured to display a correspondence between each search term and a result of the search operation, wherein the visualization unit (104, 114) displays the result obtained by the search operation at the center of a display mode and displays the search term at a specific position in the display mode, and wherein the higher the similarity of the search term to the result obtained by the search operation, the closer the specific position to the center at which the visualization unit (104, 114) displays the search term.

2. The search system (10) according to claim 1, wherein the one or more search terms include an original search term and an expanded term derived from the original search term, wherein the full-text search unit (112) performs the search operation based on the original search term and the expanded term, and wherein the visualization unit (114) displays a correspondence between each original search term and the expanded term and a result obtained by performing the search operation.

3. The search system (10) according to claim 2, wherein the expanded term is a word extracted from a result obtained by performing the search operation based on the original search term by applying pseudo-relevance feedback (PRF).

4. The search system (10) according to claim 2, wherein the expanded term is at least any one of a synonym of the original search term, a word indicating a superordinate concept of the original search term, a word indicating a subordinate concept of the original search term, and a word having a co-occurrence relationship with the original search term.

5. The search system (10) according to any one of claims 1 to 4, further comprising a search sentence correction unit (115) configured to correct a query used to perform the search operation, and perform a next search operation based on the corrected query. wherein 6. The search system (10) according to any one of claims 1 to 4, wherein the visualization unit (104, 114) displays a correspondence between the search term and the result obtained by the search operation based on a number of occurrences of the search term in the result obtained by the search operation and a similarity of each search term in the one or more search terms to the result obtained by the search operation.

7. The search system (10) according to any one of claims 1 to 4, wherein the larger the number of occurrences of the search term in the result obtained by the search operation, the larger the character size with which the visualization unit (104, 114) displays the search term.

8. The search system (10) according to any one of claims 1 to 4, wherein ​ ​ ​ ​ wherein ​ ​ wherein ​ ​ ​ ​ wherein ​ ​ ​ ​ wherein The higher the similarity of the search term to the results obtained by the search operation, the closer to the center the visualization unit (104, 114) displays the results obtained by the search operation at a certain position, the greater the number of occurrences of the search term in the results obtained by the search operation, the larger the character size the visualization unit (104, 114) uses to display the results obtained by the search operation.

9. The search system (10) of claim 7, wherein when the display mode is a spiral mode, the visualization unit (104, 114) displays the results obtained by the search operation at the center of the spiral mode and the search term along the spiral mode, or the visualization unit (104, 114) displays the search term at the center of the spiral mode and the results obtained by the search operation along the spiral mode.

10. A method comprising: receiving a search sentence; obtaining one or more search terms from the search sentence; performing a search operation according to the one or more search terms, and displaying a corresponding relationship between each search term and the results obtained by the search operation; wherein, displaying the results obtained by the search operation and the search term by a display mode, the results obtained by the search operation are displayed at the center of the display mode and the search term is displayed at a certain position in the display mode, wherein the higher the similarity of the search term to the results obtained by the search operation, the closer to the center the search term is displayed at a certain position.

11. A carrier device, storing one or more instructions that, when executed by one or more processors, cause the one or more processors to perform the method of claim 10.

Images