Information processing apparatus, generation method, and recording medium

Abstract

An information processing apparatus (100) includes an acquisition unit (120) that acquires a plurality of documents (111) and clue information (112) including a plurality of words, and a generation processing unit (130) that generates inter-document relationship raw information (200) based on the plurality of documents (111) and the clue information (112), generates high-frequency appearance information (210) based on the inter-document relationship raw information (200), generates near appearance information (220) based on the high-frequency appearance information (210), and generates an inter-document relationship table (113) based on the near appearance information (220).

Description

TECHNICAL FIELD

[0001] The present application relates to an information processing apparatus, a generation method, and a recording medium on which a generation program is recorded. BACKGROUND

[0002] A retrieval technique of a document is known. For example, a document vector is used in retrieval. Here, a technique related to a document vector has been proposed (see Patent Literature 1).

[0003] PRIOR ART DOCUMENT

[0004] PATENT LITERATURE

[0005] Patent Literature 1: Japanese Patent Application Laid-Open No. 2018-136760 SUMMARY

[0006] PROBLEMS TO BE SOLVED BY THE INVENTION

[0007] However, the apparatus retrieves a plurality of documents containing all or a part of the input keyword. However, in the existing retrieval based on a document vector, a plurality of similar documents containing the input keyword or a keyword similar to the input keyword are retrieved, but the relationality between these similar documents cannot be captured. In addition, a document having strong relevance cannot be retrieved using a viewpoint different from the input keyword. For example, in the case of retrieving a design document of a certain system using a keyword such as "warehouse performance", a document associated with "constituent by equipment division" as another viewpoint associated with the keyword is not retrieved. In this way, in order to retrieve a document having strong relevance between similar documents when similar documents are retrieved, it is necessary to associate similar documents and relevant documents in advance. For example, a method of associating documents having a relationship with all the words contained in similar documents respectively is considered. However, in this method, documents having such weak relationality that they are associated only by one of the plurality of words contained in the documents are sometimes associated with similar documents.

[0008] In addition, a plurality of words are contained in one document, and therefore, in this method, a large amount of relevance is obtained, and there is a problem that it is not known which associated document is better to refer to by the user.

[0009] An object of the present application is to prevent association with documents having weak relationality in the above-mentioned such prior association.

[0010] MEANS FOR SOLVING THE PROBLEMS

[0011] An information processing apparatus of one embodiment of the present application is provided. The information processing apparatus includes an acquisition unit that acquires a plurality of documents and clue information including a plurality of words, and a generation processing unit that generates inter-document relationship raw information indicating a relationship between a plurality of documents including at least one of the plurality of words included in the clue information and positions of the plurality of words included in the clue information in the documents, generates high-frequency appearance information indicating a relationship between a plurality of documents including the plurality of words corresponding to the number of documents equal to or greater than a predetermined threshold and positions of the plurality of words corresponding to the number of documents equal to or greater than the threshold in the documents, generates neighborhood appearance information indicating documents in which the plurality of words corresponding to the number of documents equal to or greater than the threshold appear within a predetermined range, and generates inter-document relationship information indicating a relationship between documents based on the neighborhood appearance information.

[0012] Effects of Invention

[0013] According to the present application, it is possible to prevent association with documents having weak relationships. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 FIG. 1 is a block diagram illustrating functions of an information processing apparatus of Embodiment 1.

[0015] Figure 2 FIG. 2 is a diagram illustrating hardware of the information processing apparatus of Embodiment 1.

[0016] Figure 3 FIG. 3 is a diagram illustrating an example of a generation process of inter-document relationship raw information of Embodiment 1.

[0017] Figure 4 FIG. 4 is a diagram illustrating an example of the inter-document relationship raw information of Embodiment 1.

[0018] Figure 5 FIG. 5 is a diagram illustrating an example of high-frequency appearance information of Embodiment 1.

[0019] Figure 6 FIG. 6 is a diagram illustrating an example of neighborhood appearance information of Embodiment 1.

[0020] Figure 7 FIG. 7 is a diagram illustrating an example of new inter-document relationship raw information of Embodiment 1.

[0021] Figure 8 FIG. 8 is a diagram illustrating an example of an inter-document relationship table of Embodiment 1.

[0022] Figure 9 is a flowchart showing an example of processing performed by the information processing apparatus of Embodiment 1.

[0023] Figure 10 is a diagram showing a modification example of the inter-document relationship table of Embodiment 1.

[0024] Figure 11 is a diagram showing an example of clue information of Embodiment 2.

[0025] Figure 12 is a diagram showing an example of inter-document relationship raw information of Embodiment 2.

[0026] Figure 13 is a diagram showing an example of the inter-document relationship table of Embodiment 2.

[0027] Figure 14 is a flowchart showing an example of processing performed by the information processing apparatus of Embodiment 2. DETAILED DESCRIPTION

[0028] Embodiments will be described below with reference to the accompanying drawings. The following embodiments are nothing more than examples, and various modifications can be made within the scope of the present application.

[0029] Embodiment 1

[0030] Figure 1 is a block diagram showing the functions of the information processing apparatus of Embodiment 1. The information processing apparatus 100 is an apparatus that performs a generation method. The information processing apparatus 100 has a storage section 110, an acquisition section 120, and a generation processing section 130.

[0031] Here, the hardware possessed by the information processing apparatus 100 will be described.

[0032] Figure 2 is a diagram showing the hardware possessed by the information processing apparatus of Embodiment 1. The information processing apparatus 100 has a processor 101, a volatile storage apparatus 102, and a non-volatile storage apparatus 103.

[0033] The processor 101 controls the entire information processing apparatus 100. For example, the processor 101 is a CPU (Central Processing Unit), an FPGA (Field Programmable Gate Array), or the like. The processor 101 can also be a multi-processor. Furthermore, the information processing apparatus 100 can also have a processing circuit. The processing circuit can also be a single circuit or a composite circuit.

[0034] The volatile storage device 102 is a main storage device of the information processing device 100. For example, the volatile storage device 102 is a RAM (Random Access Memory). The non-volatile storage device 103 is an auxiliary storage device of the information processing device 100. For example, the non-volatile storage device 103 is a HDD (Hard Disk Drive) or a SSD (Solid State Drive).

[0035] Returning to Figure 1 The function that the information processing device 100 has will be described.

[0036] The storage section 110 can also be realized as a storage area secured in the volatile storage device 102 or the non-volatile storage device 103.

[0037] Part or all of the acquisition section 120 and the generation processing section 130 can also be realized by a processing circuit. Further, part or all of the acquisition section 120 and the generation processing section 130 can also be realized as a module of a program executed by the processor 101. For example, the program executed by the processor 101 is also called a generation program. For example, the generation program is recorded in a recording medium.

[0038] The storage section 110 can also store a plurality of documents 111 and clue information 112. Figure 1 It is shown that the documents A, the documents B, the documents C, and the documents D are included in the plurality of documents 111. Figure 1 "Wan", "Wbn", "Wcn", and "Wdn" shown in the documents of the plurality of documents 111 are words. In addition, "n" is a positive integer. For example, Figure 1 "Wa1" and "Wa2" shown in the document A of the plurality of documents 111 are words.

[0039] The clue information 112 can also be called a clue dictionary. The clue information 112 is information indicating a clue of a relationship between documents. Figure 1 "KWn" (n is a positive integer) shown in the clue information 112 is a word. For example, "KW1", "KW2", and "KW3" shown in the clue information 112 are words. In this way, the clue information 112 includes a plurality of words. In addition, for example, the clue information 112 can also be generated by a user operating the information processing device 100.

[0040] The acquisition section 120 acquires the plurality of documents 111. For example, the acquisition section 120 acquires the plurality of documents 111 from the storage section 110. Here, the plurality of documents 111 can also be stored in an external device (for example, a cloud server). In a case where the plurality of documents 111 are stored in the external device, the acquisition section 120 acquires the plurality of documents 111 from the external device.

[0041] The acquisition unit 120 acquires the clue information 112. For example, the acquisition unit 120 acquires the clue information 112 from the storage unit 110. Here, the clue information 112 can also be stored in an external device. In the case where the clue information 112 is stored in an external device, the acquisition unit 120 acquires the clue information 112 from the external device.

[0042] The generation processing unit 130 generates inter-document relationship raw information from the plurality of documents 111 and the clue information 112. The inter-document relationship raw information indicates a relationship between the plurality of documents that contain at least one of the plurality of words contained in the clue information 112 and the occurrence positions of the plurality of words contained in the clue information 112 in the respective documents. The generation processing of the inter-document relationship raw information is described using a specific example.

[0043] Figure 3 Fig. 1 is a diagram illustrating an example of the generation processing of the inter-document relationship raw information of Embodiment 1. Figure 3 It is shown that "warehousing", "warehousing-out", and "performance" are contained in the clue information 112. Furthermore, Figure 3 It is shown that the document IDs (identifiers) "1", "5", "7", and "10". In addition, "Wn" (n is a positive integer) shown in the documents is a word. For example, "W11" is a word.

[0044] The generation processing unit 130 retrieves the documents that contain "warehousing" from the plurality of documents 111 and the clue information 112. Thereby, the documents of the document IDs "1", "5", and "10" are retrieved. Similarly, the generation processing unit 130 retrieves the documents that contain "warehousing-out" from the plurality of documents 111 and the clue information 112. Thereby, the document of the document ID "7" is retrieved. The generation processing unit 130 retrieves the documents that contain "performance" from the plurality of documents 111 and the clue information 112. Thereby, the documents of the document IDs "5" and "7" are retrieved.

[0045] In this way, for example, the generation processing unit 130 retrieves the documents of the document IDs "1", "5", "7", and "10" that contain at least one of "warehousing", "warehousing-out", and "performance" contained in the clue information 112.

[0046] The generation processing unit 130 detects the occurrence positions of the respective plurality of words contained in the clue information 112 in the documents. For example, the generation processing unit 130 detects the occurrence position of "warehousing" in the document of the document ID "1".

[0047] The generation processing unit 130 generates the inter-document relationship raw information that indicates the relationship between the retrieved documents and the occurrence positions. An example of the inter-document relationship raw information is shown.

[0048] Figure 4This is a diagram showing an example of the original document - to - document relationship information of Embodiment 1. The original document - to - document relationship information 200 has items of clue words, document IDs, occurrence lines, and occurrence character positions. Additionally, the items of occurrence lines and occurrence character positions can also be considered as items of occurrence positions.

[0049] For example, the original document - to - document relationship information 200 indicates that the document with document ID “1” contains “warehousing”. In addition, the original document - to - document relationship information 200 indicates that “warehousing” appears at the 1st character of the 2nd line of the document with document ID “1”.

[0050] In this way, the generation processing unit 130 generates the original document - to - document relationship information 200.

[0051] The generation processing unit 130 determines whether the number of documents corresponding to each of the multiple words included in the clue information 112 is greater than or equal to a preset threshold based on the original document - to - document relationship information 200. For example, the generation processing unit 130 detects the number of documents corresponding to “warehousing” included in the clue information 112 based on the original document - to - document relationship information 200. Specifically, the number of documents corresponding to “warehousing” is the number of document IDs such as document ID “1”, “5”, “10”, etc. The generation processing unit 130 determines whether the number of documents corresponding to “warehousing” is greater than or equal to the threshold based on the original document - to - document relationship information 200. Additionally, for example, this threshold is the number of documents calculated based on the number of multiple documents 111. For example, this threshold is set in a form such as 30% of the number of multiple documents 111. Similarly, the generation processing unit 130 detects the number of documents corresponding to “out - of - warehouse” and the number of documents corresponding to “actual performance”. Based on the original document - to - document relationship information 200, it is determined whether the number of documents corresponding to “out - of - warehouse” and “actual performance” is greater than or equal to this threshold.

[0052] The generation processing unit 130 generates high - frequency occurrence information based on the determination result. A specific example of the high - frequency occurrence information is shown.

[0053] Figure 5 This is a diagram showing an example of the high - frequency occurrence information of Embodiment 1. The high - frequency occurrence information 210 has items of clue words, document IDs, occurrence lines, and occurrence character positions.

[0054] The generation processing unit 130 generates the high - frequency occurrence information 210, which represents the relationship between the multiple documents containing the multiple words (i.e., “warehousing” and “actual performance”) corresponding to the number of documents greater than or equal to the threshold and the occurrence positions of each of the multiple words corresponding to the number of documents greater than or equal to the threshold in the documents.

[0055] Figure 5The number of documents corresponding to "inbound" and "performance" is shown to be equal to or greater than the threshold value. In addition, the number of documents corresponding to "outbound" is less than the threshold value, and thus, the documents including only "outbound" are not included in the high frequency appearance information 210.

[0056] The generation processing section 130 generates, based on the high frequency appearance information 210, nearby appearance information indicating documents in which a plurality of words corresponding to the number of documents equal to or greater than the threshold value appear in a document within a predetermined range. In detail, the generation processing section 130 generates, based on the high frequency appearance information 210, nearby appearance information indicating documents in which a plurality of words corresponding to the number of documents equal to or greater than the threshold value appear in a document within a predetermined range from a word. A specific example of the nearby appearance information is shown.

[0057] Figure 6 Fig. 22 is a diagram showing an example of the nearby appearance information of Embodiment 1. The nearby appearance information 220 has items of combinations of line words, document IDs, appearance lines, and appearance character positions.

[0058] For example, the nearby appearance information 220 indicates that "inbound" and "performance" corresponding to the number of documents equal to or greater than the threshold value appear in the document ID "5" of the document within a predetermined range.

[0059] In addition, for example, the predetermined range is 2 lines. In addition, for example, the predetermined range is a range of 10 characters before and after "inbound" with "inbound" as a reference.

[0060] "Inbound" and "performance" appear in the same line (i.e., the "4th" line) of the document of the document ID "5". Therefore, the document ID "5" is included in the nearby appearance information 220.

[0061] The generation processing section 130 deletes the high frequency appearance information 210 from the inter-document relationship raw information 200. The generation processing section 130 adds the nearby appearance information 220 to the inter-document relationship raw information 200, thereby generating new inter-document relationship raw information. The new inter-document relationship raw information is exemplified.

[0062] Figure 7 Fig. 23 is a diagram showing an example of the new inter-document relationship raw information of Embodiment 1. In the new inter-document relationship raw information 230, the high frequency appearance information 210 is deleted. In addition, the nearby appearance information 220 is added to the new inter-document relationship raw information 230.

[0063] The generation processing section 130 generates an inter-document relationship table based on the new inter-document relationship raw information 230. The inter-document relationship table is information indicating a relationship between documents. The inter-document relationship table is specifically shown.

[0064] Figure 8is a drawing showing an example of the inter-document relation table of Embodiment 1. The inter-document relation table 113 is also called inter-document relation information. The inter-document relation table 113 has items of combinations of line words, document IDs, and inter-document relation scores.

[0065] For example, the inter-document relation table 113 shows that the document of the document ID "5" and the document of the document ID "15" have a relation. The document of the document ID "5" and the document of the document ID "15" are documents that contain "warehouse" and "performance". Therefore, the document of the document ID "5" and the document of the document ID "15" are strongly related.

[0066] The inter-document relation score represents a degree of relation between documents. For example, the inter-document relation score is a tf-idf value. In the case where the tf-idf value is calculated, "warehouse" and "performance" can be regarded as one word, and a total value of the tf-idf value of the document ID "5" and the tf-idf value of the document ID "15" can be calculated as the inter-document relation score. Further, an average value of the tf-idf value of the document ID "5" and the tf-idf value of the document ID "15" can be calculated as the inter-document relation score.

[0067] Further, a total value of the tf-idf value of "warehouse" of the document ID "5", the tf-idf value of "performance" of the document ID "5", the tf-idf value of "warehouse" of the document ID "15", and the tf-idf value of "performance" of the document ID "15" can be calculated as the inter-document relation score. Further, an average value obtained by using the total value can be calculated as the inter-document relation score.

[0068] Further, for example, the inter-document relation score is a score of Okapi BM25. The score is represented using Equation (1).

[0069]

[0070] Further, q i is a group of high-frequency important words contained in a document D (for example, a combination of "warehouse" and "performance"). f(q i , D) is the frequency of occurrence of q i in the document D. |D| is the number of words of the document D. avgdI is the average number of words of the document set. k1 and b are arbitrary parameters.

[0071] IDF(q i ) is the IDF value of a word. IDF(q i ) is represented using Equation (2).

[0072]

[0073] N is the total number of documents. n(q i ) is the number of documents containing q i .

[0074] Next, the processing performed by the information processing apparatus 100 is described using a flowchart.

[0075] Figure 9 is a flowchart showing an example of the processing performed by the information processing apparatus of Embodiment 1.

[0076] The generation processing section 130 generates inter-document relationship raw information 200 from the plurality of documents 111 and the clue information 112 (step S11).

[0077] The generation processing section 130 generates high frequency occurrence information 210 using the inter-document relationship raw information 200 (step S12).

[0078] The generation processing section 130 generates nearby occurrence information 220 from the high frequency occurrence information 210 (step S13).

[0079] The generation processing section 130 generates new inter-document relationship raw information 230 (step S14).

[0080] The generation processing section 130 generates an inter-document relationship table 113 from the new inter-document relationship raw information 230 (step S15). In addition, the generation processing section 130 calculates an inter-document relationship score from the new inter-document relationship raw information 230. The calculated inter-document relationship score is included in the inter-document relationship table 113. The generation processing section 130 can also store the inter-document relationship table 113 in the storage section 110.

[0081] The case where the inter-document relationship table 113 is generated from the new inter-document relationship raw information 230 is described above. The generation processing section 130 can also generate the inter-document relationship table 113 from the nearby occurrence information 220. In the case where the inter-document relationship table 113 is generated from the nearby occurrence information 220, a record of "out of stock" is not generated in the inter-document relationship table 113. In addition, the generation processing section 130 can also calculate an inter-document relationship score from the nearby occurrence information 220. In detail, the generation processing section 130 calculates an inter-document relationship score from a plurality of words shown in the nearby occurrence information 220 and a plurality of documents containing the plurality of words. Then, the generation processing section 130 can also generate the inter-document relationship table 113 including the calculated inter-document relationship score. Figure 8

[0082] ​Here, an example of a method of using the inter-document relationship table 113 will be described. For example, "warehouse" is input as a keyword to the information processing apparatus 100. The information processing apparatus 100 detects a document including the document ID "5" and the like containing "warehouse" by the existing method of Patent Literature 1 or the like. The information processing apparatus 100 refers to the inter-document relationship table 113, and determines the document ID "15" and the document ID "25" having strong relationship with the document ID "5" included in the search result. The information processing apparatus 100 outputs the document of the document ID "15" and the document of the document ID "25" as the documents associated with the document ID "5" in a form different from the search result. Further, the information processing apparatus 100 determines the order of the output documents according to the inter-document relationship scores when outputting these associated documents.

[0083] In this way, the information processing apparatus 100 can output the list of the associated documents in order from strong to weak relationship between each document of the search result by using the inter-document relationship scores.

[0084] Here, as a method of associating documents, a method of associating documents having relationship with all words included in one document respectively is considered. However, in this method, documents having weak relationship by being associated with only one word of a plurality of words included in the one document are sometimes associated with the one document. Further, a plurality of words are included in one document, and thus, a large number of associations are obtained in this method, and there is a problem that a user does not know which associated document to refer to.

[0085] The information processing apparatus 100 does not associate documents having relationship with all words included in one document respectively when generating the inter-document relationship table 113. The information processing apparatus 100 associates documents including a plurality of words appearing with high frequency in the inter-document relationship raw information 200 and the plurality of words appearing in the vicinity in the document with each other. The associated documents can be said to have strong relationship with each other. The associated documents are registered in the inter-document relationship table 113. Thus, according to Embodiment 1, the information processing apparatus 100 can prevent association with documents having weak relationship according to the inter-document relationship table 113.

[0086] Here, the inter-document relationship table 113 can be changed as follows.

[0087] Figure 10 is a diagram illustrating a modification example of the inter-document relationship table of Embodiment 1. It is assumed that the inter-document relationship table 113 of Figure 10 is generated. Figure 10 The inter-document relationship table 113 of Figure 10The inter-document relationship table 113 of the document with the document ID "5" and the document with the document ID "15" indicates that the documents have a relationship. The inter-document relationship is the same, and therefore, "inbound" and "performance" can also be aggregated into 1 with "outbound". The inter-document relationship score can also be a total value of the respective inter-document relationship scores or an average value of the inter-document relationship scores.

[0088] Embodiment 2

[0089] Next, Embodiment 2 is described. In Embodiment 2, matters different from Embodiment 1 are mainly described. Also, in Embodiment 2, description of matters common to Embodiment 1 is omitted.

[0090] In Embodiment 1, a case where a plurality of words are included in the clue information 112 is described. In Embodiment 2, a case where a compound word or a sentence is included in the clue information 112 is described.

[0091] First, a case where a compound word is included in the clue information 112 is described. The clue information 112 including a compound word is exemplified.

[0092] Figure 11 is a diagram showing an example of the clue information of Embodiment 2. Figure 11 "According to the device, the constituent expansion data output is included in the clue information 112. "According to the device, the constituent expansion data output" is a compound word. "According to the device, the constituent expansion data output" is a word in which "according to the device", "constituent", "expansion", "data", and "output" (i.e., a plurality of words) are continuous.

[0093] The generation processing section 130 can extract the plurality of words (i.e., "according to the device", "constituent", "expansion", "data", and "output") constituting "according to the device, the constituent expansion data output" using morphological analysis.

[0094] The generation processing section 130 detects, from the plurality of documents 111 and the clue information 112, a plurality of documents in which the plurality of words constituting the compound word appear within a range set in advance. For example, the generation processing section 130 detects a plurality of documents (e.g., the documents with the document IDs "1" and "10") in which "according to the device", "constituent", "expansion", "data", and "output" appear within the range. Also, for example, the range set in advance is 2 lines. In addition, the order in which "according to the device", "constituent", "expansion", "data", and "output" appear within the range can be the same as the order of "according to the device", "constituent", "expansion", "data", and "output", or can be different.

[0095] The generation processing section 130 detects the appearance position of each of the plurality of words that constitute the compound word in the document.

[0096] The generation processing section 130 generates document-to-document relationship raw information that indicates the relationship between the detected document and the appearance position. An example of the document-to-document relationship raw information is shown.

[0097] Figure 12 is a diagram showing an example of the document-to-document relationship raw information of Embodiment 2. The document-to-document relationship raw information 200a has items of line word, document ID, appearance line, and appearance character position. In addition, the items of appearance line and appearance character position can be considered as items of appearance position.

[0098] For example, the document-to-document relationship raw information 200a indicates that "by device differentiation", "constituent", "expansion", "data", and "output" are included in the document of document ID "1". In addition, the document-to-document relationship raw information 200a indicates that any one of "by device differentiation", "constituent", "expansion", "data", and "output" appears in the 2nd character of the 10th line of the document of document ID "1".

[0099] In this way, the generation processing section 130 generates the document-to-document relationship raw information 200a.

[0100] The above describes a case where the plurality of documents in which all the words that constitute the compound word appear in the range are detected. The generation processing section 130 can also detect the plurality of documents in which at least one of the plurality of words that constitute the compound word appears in the range. For example, the generation processing section 130 detects the plurality of documents in which the four words of "by device differentiation", "constituent", "data", and "output" appear in the range.

[0101] Here, the information of the block 201 of the document-to-document relationship raw information 200a is not changed until the document-to-document relationship table is generated. That is, the processing after the generation of the document-to-document relationship raw information 200a is the same as Embodiment 1.

[0102] The generation processing section 130 generates a document-to-document relationship table from the detected plurality of documents. An example of the document-to-document relationship table is shown.

[0103] Figure 13 is a diagram showing an example of the document-to-document relationship table of Embodiment 2. For example, the generation processing section 130 generates the document-to-document relationship table 113a from the detected plurality of documents (for example, the documents of document IDs "1" and "10"). The document-to-document relationship table 113a indicates that the document of document ID "1" and the document of document ID "10" have a relationship.

[0104] The generation processing section 130 calculates an inter-document relation score based on at least one of the plurality of words constituting the compound word or the plurality of words constituting the compound word and the plurality of documents detected. For example, in the case of calculating the tf-idf value, the generation processing section 130 regards "distinguishing by device", "constituting", "expanding", "data", and "output" as one word, and calculates a total value of the tf-idf value of the document ID "1" and the tf-idf value of the document ID "10" as the inter-document relation score. The calculated inter-document relation score is included in the inter-document relation table 113a.

[0105] Next, a case where a sentence is included in the clue information 112 is described.

[0106] A sentence includes a plurality of words. A sentence can also be expressed as a character string including at least an auxiliary word and a plurality of words. For example, the sentence is "expanding data constituting distinguished by device".

[0107] The generation processing section 130 detects a plurality of documents in which the plurality of words included in the sentence appear within a range set in advance, based on the plurality of documents 111 and the clue information 112. In addition, for example, the range set in advance is 2 lines. Furthermore, the generation processing section 130 can extract the plurality of words included in the sentence using morphological analysis. For example, the generation processing section 130 can extract "distinguishing by device", "constituting", "expanding", "data", and "output" included in the sentence using morphological analysis. In addition, as exemplified, a word of a verb can also be changed to a word of a noun.

[0108] The generation processing section 130 detects the appearance position of each of the plurality of words included in the sentence in the document.

[0109] The generation processing section 130 generates inter-document relation raw information indicating the relation between the detected document and the appearance position. Thereby, information such as the inter-document relation raw information 200a is generated.

[0110] The above describes a case where a plurality of documents in which all the words included in the sentence appear within the range are detected. The generation processing section 130 can also detect a plurality of documents in which at least one of the plurality of words included in the sentence appears within the range.

[0111] As described above, until the inter-document relation table is generated, information indicating the relation between the plurality of words included in the sentence, the detected document, and the appearance position (for example, the information of the block 201 of Figure 12 ) is not changed.

[0112] The generation processing section 130 generates an inter-document relationship table from the detected plurality of documents. Thus, information such as the inter-document relationship table 113a is generated. Further, the inter-document relationship score included in the generated inter-document relationship table can be calculated as follows. The generation processing section 130 calculates the inter-document relationship score from at least one of the plurality of words included in the sentence or the plurality of words included in the sentence and the detected plurality of documents.

[0113] Next, the processing performed by the information processing apparatus 100 is described using a flowchart.

[0114] Figure 14 FIG. 14 is a flowchart illustrating an example of the processing performed by the information processing apparatus according to Embodiment 2.

[0115] (S21) The generation processing section 130 generates the inter-document relationship raw information 200a from the plurality of documents 111 and the clue information 112. Until the inter-document relationship table 113a is generated, the information of the block 201 of the inter-document relationship raw information 200a is not changed.

[0116] (S22) The generation processing section 130 generates the high frequency appearance information 210 using the inter-document relationship raw information 200a. In addition, in generating the high frequency appearance information 210, the generation processing section 130 does not use the information of the block 201 of the inter-document relationship raw information 200a.

[0117] (S23) The generation processing section 130 generates the nearby appearance information 220 from the high frequency appearance information 210.

[0118] (S24) The generation processing section 130 deletes the high frequency appearance information 210 from the inter-document relationship raw information 200a. The generation processing section 130 adds the nearby appearance information 220 to the inter-document relationship raw information 200a, thereby generating new inter-document relationship raw information.

[0119] (S25) The generation processing section 130 generates the inter-document relationship table 113a from the new inter-document relationship raw information. Further, the generation processing section 130 calculates the inter-document relationship score from the new inter-document relationship raw information. The calculated inter-document relationship score is included in the inter-document relationship table 113a. The generation processing section 130 can also store the inter-document relationship table 113a in the storage section 110.

[0120] Here, the number of documents containing a compound word or a sentence that is completely identical to the compound word or the sentence contained in the clue information 112 is small. Therefore, when the compound word or the sentence is contained in the clue information 112, it is difficult to detect a document that is strongly related. However, the information processing apparatus 100 executes the above-described method, whereby the information processing apparatus 100 can also detect a document that is strongly related when the compound word or the sentence is contained in the clue information 112.

[0121] Features in each of the embodiments described above can be appropriately combined with each other.

[0122] Label Explanation

[0123] 100: information processing apparatus; 101: processor; 102: volatile storage apparatus; 103: nonvolatile storage apparatus; 110: storage section; 111: plurality of documents; 112: clue information; 113, 113a: inter-document relationship table; 120: acquisition section; 130: generation processing section; 200, 200a: inter-document relationship raw information; 201: frame; 210: high frequency occurrence information; 220: nearby occurrence information; 230: new inter-document relationship raw information.

Claims

1. An information processing apparatus, the information processing apparatus comprising: an acquisition unit that acquires a plurality of documents and clue information including a plurality of words; and a generation processing unit that generates, based on the plurality of documents and the clue information, document-to-document relationship raw information indicating a relationship between a plurality of documents including at least one of the plurality of words included in the clue information and positions in the documents at which the plurality of words included in the clue information each appear, determines, based on the document-to-document relationship raw information, whether a number of documents corresponding to the plurality of words included in the clue information is equal to or greater than a threshold value set in advance, generates high-frequency appearance information indicating a relationship between a plurality of documents including the plurality of words corresponding to the number of documents equal to or greater than the threshold value and positions in the documents at which the plurality of words corresponding to the number of documents equal to or greater than the threshold value each appear, generates, based on the high-frequency appearance information, nearby appearance information indicating documents in which the plurality of words corresponding to the number of documents equal to or greater than the threshold value appear within a range set in advance, and generates document-to-document relationship information indicating a relationship between documents based on the nearby appearance information. 2.The information processing apparatus according to claim 1, wherein the generation processing unit calculates, based on the nearby appearance information, a document-to-document relationship score indicating a degree of a relationship between documents, and generates the document-to-document relationship information including the document-to-document relationship score. 3.The information processing apparatus according to claim 1, wherein the clue information includes a word composed of a plurality of words in succession, i.e., a compound word, the generation processing unit detects, based on the plurality of documents and the clue information, a plurality of documents in which at least one of the plurality of words constituting the compound word appears within a range set in advance, and generates the document-to-document relationship information based on the detected plurality of documents. 4.The information processing apparatus according to claim 3, wherein the generation processing unit detects, based on the plurality of documents and the clue information, a plurality of documents in which the plurality of words constituting the compound word appear within a range set in advance, and generates the document-to-document relationship information based on the detected plurality of documents. 5.The information processing apparatus according to claim 3 or 4, wherein the generation processing unit calculates, based on at least one of the plurality of words constituting the compound word or the plurality of words constituting the compound word and the detected plurality of documents, a document-to-document relationship score indicating a degree of a relationship between documents, and generates the document-to-document relationship information including the document-to-document relationship score. 6.The information processing apparatus according to claim 1, wherein the clue information includes a sentence including a plurality of words, the generation processing unit detects, based on the plurality of documents and the clue information, a plurality of documents in which at least one of the plurality of words included in the sentence appears within a range set in advance, and generates the document-to-document relationship information based on the detected plurality of documents. 7.The information processing apparatus according to claim 6, wherein The generation processing section detects, from the plurality of documents and the clue information, a plurality of documents in which a plurality of words included in the sentence appear within a predetermined range, and generates the inter-document relationship information based on the detected plurality of documents.

8. The information processing apparatus according to claim 6 or 7, wherein The generation processing section calculates an inter-document relationship score indicating a degree of relationship between documents based on at least one of a plurality of words included in the sentence or the plurality of words included in the sentence and the detected plurality of documents, and generates the inter-document relationship information including the inter-document relationship score.

9. A generation method, wherein An information processing apparatus acquires a plurality of documents and clue information including a plurality of words, The information processing apparatus generates, based on the plurality of documents and the clue information, inter-document relationship raw information indicating a relationship between a plurality of documents including at least one of a plurality of words included in the clue information and positions in the documents at which the plurality of words included in the clue information each appear, The information processing apparatus determines, based on the inter-document relationship raw information, whether a number of documents corresponding to the plurality of words included in the clue information is equal to or greater than a predetermined threshold value, The information processing apparatus generates high-frequency appearance information indicating a relationship between a plurality of documents including the plurality of words corresponding to the number of documents equal to or greater than the threshold value and positions in the documents at which the plurality of words corresponding to the number of documents equal to or greater than the threshold value each appear, The information processing apparatus generates, based on the high-frequency appearance information, nearby appearance information indicating documents in which the plurality of words corresponding to the number of documents equal to or greater than the threshold value appear within a predetermined range, The information processing apparatus generates, based on the nearby appearance information, inter-document relationship information indicating a relationship between documents.

10. A recording medium having a generation program recorded thereon, the generation program causing an information processing apparatus to perform the following processing: Acquire a plurality of documents and clue information including a plurality of words, Generate, based on the plurality of documents and the clue information, inter-document relationship raw information indicating a relationship between a plurality of documents including at least one of a plurality of words included in the clue information and positions in the documents at which the plurality of words included in the clue information each appear, Determine, based on the inter-document relationship raw information, whether a number of documents corresponding to the plurality of words included in the clue information is equal to or greater than a predetermined threshold value, Generate high-frequency appearance information indicating a relationship between a plurality of documents including the plurality of words corresponding to the number of documents equal to or greater than the threshold value and positions in the documents at which the plurality of words corresponding to the number of documents equal to or greater than the threshold value each appear, Generate, based on the high-frequency appearance information, nearby appearance information indicating documents in which the plurality of words corresponding to the number of documents equal to or greater than the threshold value appear within a predetermined range, Generate, based on the nearby appearance information, inter-document relationship information indicating a relationship between documents.

Images