JP6613833B2 - Information processing apparatus, information processing system, and program - Google Patents

Description

  The present invention relates to an information processing apparatus, an information processing system, and a program, and more particularly to acquisition / presentation of information using a link structure between documents.

  In recent years, with the widespread use of the Internet, users can easily search and acquire various information. For example, a Web page including a keyword input by a user can be searched and browsed by a search engine of a Web site on the Internet. In the case of hypertext with embedded links, the user can jump to a related link destination while browsing a Web page and easily obtain other information.

  In addition, for example, Patent Literature 1 describes a technique of presenting an unexpected sentence as a search result in order to display a sentence that attracts more user interest as a search result. Specifically, in the method of Patent Document 1, a document is classified into categories, and for each word extracted from the document, a word score that is an index of word unexpectedness is calculated based on the appearance frequency of each word in the category. In addition, it is described that a sentence score that is a score of an unexpectedness of a sentence is calculated based on a word score for a sentence extracted from a document. Thereby, an unexpected sentence can be extracted from a document on the assumption that a word having a high word score in a category and a high rarity are highly unexpected.

JP 2011-95905 A

  However, the above-described method of Patent Document 1 calculates the unexpectedness based on the appearance frequency of words in consideration of categories. For this reason, there is a problem in that an unexpected index cannot be calculated correctly when many of the same contents appear in different categories.

  The present invention has been made in view of such problems, and acquires and presents surprising information that is difficult for a user to find by himself / herself from related documents that are generally recognized, regardless of category. It is an object to provide an information processing apparatus, an information processing system, and a program capable of performing the above.

  According to a first aspect of the present invention, there is provided an input means for inputting a keyword, a storage means for storing a word / phrase associated with a link destination as a link relation and a document related to the phrase, and an input keyword / link. Link phrase acquisition means for acquiring a related phrase, document extraction means for extracting a document related to the input keyword and a phrase related to the input keyword, a document related to the keyword, and the link An information processing apparatus comprising: an operation unit that calculates a degree of unexpectedness between documents using a similarity to a document related to a related phrase.

  According to the first invention, the information processing apparatus includes a storage unit that stores a word / phrase associated with a link destination as a link relationship and a document related to the word / phrase, and acquires a word / phrase having a link relationship with the input keyword. , Extracting the document related to the input keyword and the document related to the phrase related to the input keyword, and using the similarity between the document related to the keyword and the document related to the phrase related to the link Calculate the degree of surprise. This makes it easier for the user to find unexpected information that is difficult to find by normal text browsing. It becomes possible to discover comprehensive information, promote AI automatic dialogue, and support ideas.

  In the first invention, it is desirable to provide an output means for outputting at least a part of a document related to the phrase having the link relation in accordance with a calculation result of the calculating means. Thereby, an unexpected part can be extracted and output from a huge amount of related documents. The user can efficiently obtain unexpected information.

  In the first invention, the storage means further includes associative word acquisition means for associating and storing words and phrases that are associated with each other, and acquiring phrases that are associated with the input keyword as associated words, It is desirable that the arithmetic means further calculates an unexpected degree of the document relating to the associative word by using the similarity between the document and the document relating to the phrase acquired by the link phrase acquiring means. As a result, an unexpected word (dissimilarity) between an input keyword and a phrase that has a link relationship with an associated word that is expected to have a high similarity is included in the input keyword group. The extraction accuracy can be improved.

  In the first invention, the storage means stores a total number referred to for each document, and the calculation means includes a document related to the input keyword and a document related to the phrase acquired by the link phrase acquisition means. It is desirable to weight the dissimilarity between documents based on the number of document references to obtain the unexpectedness. As a result, it is possible to extract surprising information from a document that is generally recognized more, and to output information that makes the user more interested.

  In the first invention, the storage means stores a phrase in association with each category, the calculation means calculates the unexpectedness for each category of the phrase, and the output means sets the inputted keyword to the keyword. On the other hand, it is desirable to output at least a part of a document relating to a phrase whose category is unexpected. As a result, the unexpectedness for each category can be calculated and used when presenting information, and information more surprising to the user can be presented.

  According to a second aspect of the present invention, there is provided a server having storage means for storing a phrase associated with a link destination as a link relation and a document relating to the phrase, an input means for inputting a keyword, and a phrase having a link relation with the input keyword. Link phrase acquisition means acquired from a server, document extraction means for extracting a document related to the input keyword and a phrase related to the input keyword from the server, a document related to the keyword, and the link relation An information processing system comprising: an information processing apparatus that includes a calculation unit that calculates an unexpected degree between documents using a similarity to a document related to a certain phrase.

  According to a second aspect of the present invention, a server having storage means for storing a phrase associated with a link destination as a link relation and a document related to the phrase is provided, and the information processing apparatus receives a phrase related to the input keyword from the server. Obtaining and extracting a document related to the input keyword and a document related to a phrase related to the input keyword from a server, and using similarity between the document related to the keyword and a document related to the phrase related to the link To calculate the unexpectedness between documents. This makes it easier for the user to find unexpected information that is difficult to find by normal text browsing. It can be used for comprehensive information discovery, AI automatic dialogue promotion, idea support, and the like.

  A third invention is a program for causing a computer to function as the information processing apparatus of the first invention. According to the third invention, it is possible to cause a computer to function as the information processing apparatus according to the first invention.

  According to the present invention, an information processing apparatus, an information processing system, and an information processing apparatus capable of acquiring and presenting unexpected information that is difficult for a user to find out of related documents that are generally recognized regardless of the category, and Can provide a program.

System configuration diagram of information processing system 1 Hardware configuration diagram of the information processing apparatus 2 The figure explaining document DB3 The figure explaining link structure DB4 An example of the data contents of the association word DB5 The figure explaining the production | generation of associative word DB5 The flowchart explaining the flow of the information acquisition process which the information processing apparatus 2 performs The figure explaining the input keyword group 52 Example of calculating dissimilarity (unexpectedness) Example of calculating similarity using document vectors Diagram explaining the output of unexpected information Diagram explaining the calculation of the unexpectedness based on the number of references Illustration explaining the unexpectedness of categories

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.

[First Embodiment]
First, a first embodiment of the present invention will be described. FIG. 1 is a diagram showing a system configuration of an information processing system 1 using an information processing apparatus 2 according to the present invention. In the information processing system 1, the information processing apparatus 2 is communicatively connected to a document DB 3, a link structure DB 4, an associative word DB 5, and the like via a network 10. The network 10 includes a LAN (Local Area Network), a WAN (Wide Area Network) connected to a wider area, or a public communication line such as the Internet, a base station, and the like. The communication connection in the network 10 may be wired or wireless.

  The information processing apparatus 2 is an apparatus such as a computer that can install an arbitrary application program and execute processing. For example, a smart phone, a tablet, a game machine, and other various information terminals are included.

  FIG. 2 is a diagram illustrating an example of a hardware configuration of the information processing apparatus 2. As shown in FIG. 2, the information processing apparatus 2 includes a control unit 21, a storage unit 22, an input unit 23, a display unit 24, a communication I / F 25, a media input / output unit 26, a peripheral device I / F unit 27, and the like. 29 is connected.

  The control unit 21 includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The CPU calls a program stored in the storage unit 22, ROM, recording medium or the like to a work memory area on the RAM and executes it, and drives and controls each unit connected via the bus 29.

  The CPU of the control unit 21 executes an information acquisition process that acquires and outputs information that is surprising for an input arbitrary keyword. Details of this information acquisition processing will be described later.

  The ROM permanently stores programs such as a boot program and BIOS for the information processing apparatus 2, data, and the like. The RAM temporarily holds the loaded program and data, and includes a work area used by the control unit 21 for performing various processes.

  The storage unit 22 stores a program executed by the control unit 21, data necessary for program execution, an OS (operating system), and the like. These program codes are read by the control unit 21 as necessary, transferred to the RAM, and read and executed by the CPU.

The input unit 23 is an input device such as a keyboard, a mouse, or a touch panel, for example, and outputs input data to the control unit 21.
The display unit 24 includes a display device such as a liquid crystal panel and a logic circuit for executing display processing in cooperation with the display device, and displays display information input by the control of the control unit 21 on the display device. Let In addition, when the input part 23 is comprised with a touch panel, a touch panel is comprised integrally with the display of the display part 24. FIG.

The communication I / F (interface) 25 includes an antenna, a communication control circuit, and the like, and is an interface that mediates communication with the network 10.
The media input / output unit 26 is a media input / output device such as a CD drive or a DVD drive, for example.

The peripheral device I / F (interface) unit 27 is a port for connecting a peripheral device, and transmits and receives data to and from the peripheral device via the peripheral device I / F unit 27. The connection form with the peripheral device may be wired or wireless.
The bus 29 is a path that mediates transmission / reception of control signals, data signals, and the like.

  The document DB 3 is a database that stores a document group (a plurality of documents) as shown in FIG. 3A. For example, the document DB 3 is stored in a storage area such as a Web server on the network 10 as shown in FIG. In this specification, “document” means text data in which an explanation or the like is described for a certain item. It is assumed that links (hyperlinks) to other items are set for some words included in the document. The “document group” is a collection of a plurality of the documents described above. For example, the free encyclopedia “Wikipedia” is called a document group, and each article is called a document. The document group is not limited to Wikipedia, and other dictionaries, terminology dictionaries, information articles, news articles, and the like may be included in the document group.

 As shown in FIG. 3B, the document DB 3 includes words such as “chocolate” as heading items, and each article “chocolate” is fermented and roasted with cacao seeds. The cocoa mass is used as the main ingredient, and "..." is stored as the text.

  The link structure DB 4 is a database that stores link structure data of document groups stored in the document DB 3. The link structure data is data obtained by extracting a link relation between documents in which links are embedded, and link destinations and link source words set for a certain word are collected and stored. A plurality of link destinations and link sources may be set for one word. A link is a hyperlink embedded in a document, and is information for moving to another location of the same document, another document, or a completely different site.

  For example, as shown in FIG. 4, the word “ice cream” is linked to articles (documents) related to “Natsume Soseki”, “Yamato Battleship”, “Chocolate”, “Vanilla”, and “Crepe” as link sources. Embedded. In other words, links to “Ice Cream” are embedded in articles about “Natsume Soseki”, “Yamato Battleship”, “Chocolate”, “Vanilla”, and “Crepe”. In addition, a link for jumping to “vanilla”, “crepe” or the like is embedded as a link destination in the document “ice cream”. In this case, “vanilla” and “crepe” are linked bidirectionally with “ice cream”. “Natsume Soseki”, “Yamato Battleship” and “Chocolate” are one-way links.

  As such a link structure DB 4, for example, there is “DBPedia” in which the link structures of “Wikipedia” are aggregated. “DBPedia” is a community project that is open to the public as LOD (Linked Open Data) obtained by extracting information from “Wikipedia”. The link structure data used in the present invention is not limited to “DBPedia”, and other various LODs may be used.

  The associative word DB 5 stores a plurality of associative words for each word. As shown in FIG. 5, for example, in the case of an association word of the word “ice cream”, “pudding”, “cake”, “caramel”, and the like are stored in the association word DB 5. An associative word for a word is obtained based on the number of co-occurrence words included in various document groups.

  FIG. 6 is a diagram for explaining an example of a method for generating the associative word DB 5. As shown in FIG. 6, the number of times of co-occurrence of a plurality of words extracted from various various documents is counted, and a word having the number of times of co-occurrence exceeding a preset threshold is set as an associative word. For example, as shown in the co-occurrence number counter 51 of FIG. 6, the words that appear in the same document as the word “ice cream” are “pudding” 100 times, “cake” 80 times, “caramel” 70 times, “ “Yokohama” is counted as 20 times. Of these words, “pudding”, “cake”, “caramel”, and the like that appear more frequently than the threshold are stored in the associative word DB 5 as associative words of “ice cream”. The document that is the source of the associative word DB 5 is a document such as a general Web site in addition to a Wikipedia article.

  Next, the flow of information acquisition processing in the information processing system 1 will be described. FIG. 7 is a flowchart showing a flow of information acquisition processing executed by the information processing apparatus 2. As shown in the figure, when an arbitrary keyword is input to the information processing apparatus 2 by the user (step S1), the control unit 21 of the information processing apparatus 2 acquires link structure data of the input keyword (step S2). The link structure data is stored in the link structure DB4. For example, when the input keyword is “ice cream”, the control unit 21 acquires link structure data related to “ice cream” from the link structure DB 4. As shown in FIG. 4, the link structure data related to “Ice Cream” includes “Natsume Soseki”, “Yamato Battleship”, “Chocolate”, “Vanilla”, “Crepe” as the link source, “Vanilla” as the link destination, "Crepes" are tied. The control unit 21 acquires these link source and link destination words “Natsume Soseki”, “Yamato-type battleship”, “chocolate”, “vanilla”, and “crepe” and stores them in the RAM.

  Subsequently, the control unit 21 acquires the associative word of the keyword “ice cream” input in step S1 from the associative word DB 5 (step S3). As shown in FIG. 8, it is assumed that “pudding”, “cake”, “caramel”, and the like are stored in the associative word DB 5 as associative words of the keyword “ice cream”. The control unit 21 acquires “pudding”, “cake”, and “caramel” as associative words of the input keyword “ice cream”, and stores them in the RAM as the input keyword group 52 together with the input keywords.

  The reason why an associative word is added to the input keyword group 52 is that a word or phrase that is expected to be easily associated is not surprising, so the score of the unexpectedness (dissimilarity) for such an associative word is lowered, and the accuracy of the output result This is for improvement. Therefore, when importance is attached to simplification of the system and reduction of the calculation amount rather than the accuracy of the output result, the step S3 may be omitted.

  Subsequently, the control unit 21 calculates an unexpected degree (dissimilarity) between the document of the input keyword and the associative word (input keyword group 52) and the document of the phrase having the link relationship acquired in step S2 (step S4). ). Here, a method of calculating the unexpectedness will be described with reference to FIGS. 9 and 10. FIG. 9 is a diagram for explaining the calculation of the unexpectedness (dissimilarity), and FIG. 10 is a diagram illustrating a calculation example of the similarity.

  As shown in FIG. 9, the control unit 21 firstly processes documents 7a, 7b, 7c,... For each word in the input keyword group 52 and words having a link relationship obtained from the link structure data acquired in step S2. Documents 8a,... Are read from the document DB 3, and the similarity between these documents is calculated. The similarity can be calculated using, for example, cosine similarity.

  As shown in FIG. 10, the cosine similarity is calculated by Expression (1) using document vectors A and B that are indexes representing the characteristics of each document.

  The document vector is generated based on the appearance frequency of words extracted from the document. For example, the document vector A is TF of each word based on the appearance frequency of each word “ice cream”, “milk” “raw material”, “air”,... Extracted from the document 7a regarding the input keyword “ice cream”. IDF values (1.2, 2.0, 1.25, 0.55,...) Are obtained, and these values are vectors. Note that a TF value may be used instead of the TF-IDF value. Similarly, the document vector B includes each word based on the appearance frequency of each word “chocolate”, “cacao”, “seed”, “fermentation”,... TF-IDF values (1.4, 1.5, 0.6, 0.2,...) Are obtained, and these values are vectors. The elements of document vector A and document vector B are aligned so that the same extracted words are in the same order in the vector.

As shown in FIG. 9, the control unit 21 selects the “ice cream” document 7a, the “pudding” (associative word) document 7b, the “cake” (associative word) document 7c,. Each document vector is obtained, and the similarity between the word “chocolate” and the document vector B of the document 8a having a link relationship is calculated. Further, an average value of the calculated similarities is obtained. The average of the similarities is subtracted from the numerical value “1” to obtain the dissimilarity.
Thereby, the dissimilarity with the word “chocolate” linked to the input keyword group 52 is obtained.

  Similarly, the dissimilarity with the input keyword group 52 is also obtained for the words “Natsume Soseki”, “Yamato type battleship”, “vanilla”, “crepe”,. In the first embodiment, the dissimilarity itself is regarded as an unexpected degree.

  The control unit 21 obtains some words having a high degree of surprise with the input keyword group 52 by the calculation in step S4. And the information regarding a word with a high unexpected word is acquired (step S5), and it outputs (step S6).

  As shown in FIG. 11, it is assumed that the degree of surprise is obtained for each word that has a link relationship with the input keyword group 52. For example, “Natsume Soseki” has an unexpected degree of 0.8, “Ogasawara Islands” has an unexpected degree of 0.64, “Tanaka Kakuei” has an unexpected degree of 0.5, “Abe Shinzo” has an unexpected degree of 0.4, Suppose that the unexpectedness with the input keyword group 52 (such as “chocolate”) is calculated as shown in FIG. The control unit 21 acquires one or more words having a higher unexpectedness score than a preset threshold value or a word having a higher unexpectedness score, and the document 9 relating to these words having a higher unexpectedness is obtained from the document DB 3. get. In addition, from the document 9 about words with a high degree of surprise, a sentence containing the input keyword “ice cream”, “The Great Sweet Party, I wanted the ice cream that was valuable at the time of medical treatment, and had troubled the surroundings. . "Is extracted.

  The control unit 21 outputs the extracted sentence as information on a word having a high degree of surprise. The output includes display on the display unit 24, storage in the storage unit 22, and the like. Further, printing to a printing apparatus connected to the peripheral device I / F unit 27, voice output from a voice output unit (not shown), or the like may be performed. At the time of output, the control unit 21 outputs information on a word having a high degree of surprise according to a predetermined output format. For example, as shown in FIG. 11, “Niceme Soseki” is “a big sweet party, and during treatment, I wanted an ice cream that was a valuable item at the time, and sometimes confused my surroundings. Like “It seems”, it synthesizes unexpected information into a specified output format and outputs it.

  As described above, when an arbitrary keyword is input via the input unit 23, the control unit 21 converts the phrase having a link relationship with the input keyword into the link structure data of the document group. And the unexpectedness between the document 7a related to the input keyword and the document 8a related to the link-related phrase is calculated, and the unexpectedness about the input keyword is calculated based on the calculated unexpectedness. Get and output sexual information. This makes it easier for the user to find unexpected information that is difficult to find by normal document search or browsing. It can be used for comprehensive information discovery, AI automatic dialogue promotion, idea support, and the like.

  Moreover, since the control part 21 acquires the associative word of the input keyword and calculates the unexpected degree between documents with the document of the word which has a link relation also about the document regarding an associative word, an association is anticipated easily. The word unexpectedness (dissimilarity) score can be reliably lowered, and an output result can be obtained with high accuracy.

  In the above description, as shown in FIG. 1, the information processing apparatus 2 acquires documents, link structure data, associative words, and the like from the document DB 3, the link structure DB 4, and the associative word DB 5 that are communicatively connected via the network 10. However, the present invention is not limited to this. The document DB 3, the link structure DB 4, and the associative word DB 5 are stored in the storage unit 22 in the information processing apparatus 2, and the control unit 21 of the information processing apparatus 2 acquires document data, link structure data, associative words, and the like from the storage unit 22. It is good also as composition to do.

  In addition, the link structure data includes a phrase that is a relation of a bidirectional link that is a link source and a link destination and a unidirectional link phrase that is a link source or a link destination data. You may estimate that it is a relationship similar to each other, and you may exclude from the calculation object of an unexpected degree.

[Second Embodiment]
Next, a second embodiment of the present invention will be described. In the second embodiment, when the information processing apparatus 2 calculates the unexpectedness, the unexpectedness is calculated in consideration of the number of document references in addition to the dissimilarity described in the first embodiment. The reference number is the number of times a document is referred to and can be obtained from an access log to the corresponding article (document). The reference number may be acquired via the network 10 whenever the information acquisition process is executed, or may be stored in advance in the storage unit 22 in the information processing apparatus 2 as a reference number.

  As shown in the unexpectedness calculation table 6 of FIG. 12, in the second embodiment, a value obtained by multiplying the dissimilarity by the reference number (normalized reference number) is set as the unexpectedness. The dissimilarity is obtained by the same procedure as in the first embodiment. The normalized reference number is a value (numerical value of 0 to 1) obtained by normalizing the number of word references with the total number of references.

  For example, when the dissimilarity of the word “chocolate” is “0.2” and the normalized reference number is “0.7” for the input keyword group 52, the unexpectedness is “0.14”. Also, if the dissimilarity of the word “Natsume Soseki” is “0.8” and the normalized reference number is “0.8”, the unexpectedness is “0.64”, and the word “Yamato-type battleship” When the similarity is “0.9” and the normalized reference number is “0.3”, the unexpectedness is “0.27”.

  The control unit 21 obtains the dissimilarity between the document for the input keyword group 52 and the document for the word having a link relationship in the same procedure as in the first embodiment, and the link relationship is found in the obtained dissimilarity. Multiply the normalized number of documents for a word to find the degree of surprise. After that, as in the first embodiment, information about words with a high degree of unexpectedness is acquired from the document group (document DB 3 or the like) and output.

  As described above, the information processing apparatus 2 according to the second embodiment calculates the word unexpectedness in consideration of the number of document references. By considering the number of references, it is possible to extract surprising information from more recognized words, and to attract more users' interest.

[Third Embodiment]
Next, a third embodiment of the present invention will be described. In the third embodiment, when the information processing device 2 calculates the unexpectedness of each word, the unexpectedness is calculated for each category of words and the information is output in consideration of the category.

  For example, as illustrated in FIG. 13, the control unit 21 acquires words / phrases having a link relationship with the input keyword from the link structure data, and classifies the acquired words / phrases by category. Information such as which category a word belongs to is previously given to the word or category. Moreover, the unexpectedness from the input keyword group 52 of each phrase is calculated | required, the average of unexpectedness is calculated | required for every category, and it is set as the unexpectedness of a category. Then, for example, information about a word with a high degree of surprise is acquired from a category with a high degree of unexpectedness, and is output.

  FIG. 13 is a diagram illustrating a state in which words having a link relationship with the input keyword are classified into (a) category “confectionery”, (b) category “person”, and (c) category “place”. As shown in FIG. 13A, the category “confectionery” includes “chocolate”, “puff cream”, “cheese cake”, and the like. As shown in FIG. 13B, the category “people” includes “Natsume Soseki”, “Tanaka Kakuei”, “Abe Shinzo”, and the like. As shown in FIG. 13C, the category “place” includes “Ogasawara Islands”, “Utah State University”, “Tamachi Station”, and the like.

  Then, the degree of surprise with the input keyword group 52 is obtained for each word. For example, as shown in FIG. 13A, the word “chocolate” included in the category “confectionery” has an unexpected degree of 0.14, the word “puff cream” has an unexpected degree of 0.3, and the word “cheese”. The unexpectedness of “cake” is 0.3. The unexpectedness of each word included in the other category “confectionery” is obtained, and the unexpectedness of the category is obtained as “0.3” from the average value of the unexpectedness.

  Further, as shown in FIG. 13B, the unexpected degree of the category “person” is obtained as “0.6”, and as shown in FIG. 13C, the unexpected degree of the category “place” is “0.5”. " In this case, since the category “person” in FIG. 13B has the highest degree of unexpectedness among these categories, the control unit 21 relates to the word “Natsume Soseki” having a high degree of surprise from the category “person”. Is output as unexpected information.

  Alternatively, the user may be able to select a category. In this case, the control unit 21 displays a category selection screen or the like on the display unit 24 as an interface for the user to select a category. When the user selects a category, the control unit 21 acquires information on a word / phrase having a high degree of surprise in the category selected by the user, and outputs the information as unexpected information.

  As described above, the information processing apparatus 2 according to the third embodiment further calculates the unexpectedness for each category in the information processing apparatus 2 according to the first or second embodiment and presents the information. Use. This makes it possible to present information that is more surprising to the user.

  The preferred embodiments of the information processing apparatus, the information processing system, and the program according to the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to such examples. It will be apparent to those skilled in the art that various changes or modifications can be conceived within the scope of the technical idea disclosed in the present application, and these are naturally within the technical scope of the present invention. Understood.

1 …………………… Information Processing System 2 …………………… Information Processing Device 21 …………………… Control Unit 22 …………………… Storage Unit 23 ……………… …… Input unit 24 …………………… Display unit 25 ………………… Communication I / F
26 …………………… Media Input / Output Unit 27 ………………… Peripheral Device I / F Unit 29 ………………… Bus 3 …………………… Document DB
4 …………………… Link structure DB
5 …………………… Associative Word DB
51 .............. Co-occurrence counter 6 .......... Unexpectedness calculation table 52 ................ Input keyword group 7a, 7b, 7c..Document 8a .............. ... Document 9 …………………… Document 10 about words with high degree of surprise …………………… Network

Claims (7)

An input means for inputting a keyword;
Storage means for storing a phrase associated with a link destination as a link relation and a document related to the phrase;
Link phrase acquisition means for acquiring a phrase that is linked to the input keyword,
Document extracting means for extracting a document related to the input keyword and a document related to a phrase linked to the input keyword;
Computing means for calculating the degree of surprise between documents using the similarity between the document related to the keyword and the document related to the phrase related to the link;
An information processing apparatus comprising:   The information processing apparatus according to claim 1, further comprising: an output unit that outputs at least a part of a document related to the phrase having the link relationship according to a calculation result of the calculation unit. The storage means stores words associated with each other in association with each other,
An associative word acquisition means for acquiring, as an associative word, a phrase associated with the input keyword;
The calculation means further calculates an unexpected degree of a document related to the associative word using a similarity between documents with a document related to the phrase acquired by the link phrase acquisition means. Item 3. The information processing device according to Item 2. The storage means stores the total number referred to for each document;
The calculation means weights the dissimilarity between the document related to the input keyword and the document related to the phrase acquired by the link phrase acquisition means based on the number of references of the document to obtain the unexpected degree. The information processing apparatus according to claim 1, wherein the information processing apparatus is an information processing apparatus. The storage means stores a phrase in association with each category,
The computing means calculates the unexpectedness for each category of words,
The information processing apparatus according to claim 2, wherein the output unit outputs at least a part of a document related to a phrase whose category is unexpected for the input keyword. A server having storage means for storing a phrase associated with a link destination as a link relation and a document related to the phrase;
An input means for inputting a keyword;
Link phrase acquisition means for acquiring a phrase linked to the input keyword from the server,
Document extracting means for extracting from the server a document related to the input keyword and a document related to a phrase linked to the input keyword;
An information processing apparatus comprising: an operation unit that calculates a degree of unexpectedness between documents using a similarity between the document related to the keyword and the document related to the phrase related to the link;
An information processing system comprising:   A program for causing a computer to function as the information processing apparatus according to claim 1.

Images