JP2011008599A - Item information processing apparatus for collective intelligence database system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To appropriately determine and report the necessity of editing an item to a user in processing of information related to items provided from a plurality of users through a network and registered in a collective intelligence database system such as Wikipedia.SOLUTION: With respect to a plurality of items within the collective intelligence database system, a term in a sentence and link destination information are extracted from information related to each item as attributes of the item. The extracted attributes of each item are used to calculate a similarity between two items. According to the calculated similarity, it is reported to promote integration of information related to the two items. In item division processing, with respect to one item, a sentence that is information related to the item is divided on a paragraph to paragraph basis, a similarity for each paragraph is calculated using the attributes, and cluster analysis of the paragraphs is performed using the similarity. When the cluster number is larger than one, it is reported to promote cluster-based division of the item.

Description

  The present invention relates to an item information processing apparatus that processes items provided from a plurality of users through a network and registered in a collective intelligence database system.

  Today, a collection that gathers information held by a large number of users through an electronic network such as the Internet, and aims to make the collection of information more reliable by complementing the missing information in the database and correcting errors. There is an intelligent database system. An example is “Wikipedia”.

  The actual object described in the collective intelligence database system is called an “object”. In the collective intelligence database system, the user himself / herself provides link address information on a specific site through the network or on a site where the link destination can be traced from this site, with respect to the information related to the “object” held by the information provider user. You can describe the included text. The written text and link address information are appropriately edited by various users. The text and link address information described for this “object” are provided as one item. By making one “object” correspond to one item, the collective intelligence database system becomes a system that is easy for the user to use.

  However, since the collective intelligence database system is an aggregate that gathers the knowledge of many users, it is difficult for users to use as follows.

(1) The same item is provided for a plurality of different “objects”.
For example, when there is a person who has an “object” and there is a person with the same name and the same name, information on different persons may be overlapped in the same item.
Alternatively, one item may be created even though “objects” are different from each other due to incomplete understanding of a user who provides information.

(2) A plurality of items are provided in one “object”.
For example, if a person has an “object” and the person is a Japanese person, even if the person is the same, the item name is written in alphabet, kana, kanji, etc. Provided. In addition, when a foreign word or the like is written in kana for the item name, the item name may be different depending on the presence or absence of a long sound symbol.
Or, due to incomplete understanding of the user who provides information, two items are provided even though they are actually the same “object”, or separate items are provided even though they are merely aliases. Depending on the difference or presence or absence of fine particles, new items may be provided for the same “object” without sufficiently examining already registered items. In this case, item “B” is created in addition to item “A”. That is, the information of the same “object” is dispersed into two items.
Alternatively, the name of the person should be described in the native language of the person, or the name of the person should use the notation of the place where it is used, etc. In the case of setting, even though they are the same person, one user may set the name of the person as “A”, and the other user may set the name of the person as “A ′” and newly create an item.

  As described above, even when the “object” is the same, a plurality of items are provided, or different “object” information is described as information about one item even though the “object” is different. There is.

  On the other hand, regarding multimedia, in particular, a technique using a data information source such as Wikipedia generated by a community to generate multimedia metadata is known. Specifically, first, a web page related to specific content is received from a website generated by a community such as Wikipedia. Next, a plurality of terms are extracted from the web page. Add multiple extracted terms to content metadata related to content. Thereafter, specific category data is extracted from the content metadata, the specific category data is taken into the category data set, and the number of dimensions of the category data set based on the category data set and the relation data is reduced. Finally, the relationship data defines a relationship between the category data set and content related to the category data set. In this way, multimedia metadata can be generated using a data information source such as Wikipedia generated by the community.

JP 2008-4080 A

  However, since the content metadata in the above-described known method is generated based on a data information source (collective intelligence database system) such as Wikipedia, high-quality metadata can be used unless the data information source such as Wikipedia is edited correctly. It cannot be generated.

  Therefore, the present invention provides item information that can accurately determine whether or not an item needs to be edited when editing an item that is provided from a plurality of users through a network and is registered in a collective intelligence database system such as Wikipedia. An object is to provide a processing apparatus.

The above object can be realized by an item information processing apparatus shown below that processes items provided from a plurality of users through a network and registered in the collective intelligence database system.
that time,
(A) The information related to each of the plurality of items registered in the collective intelligence database system includes a sentence explaining the contents of each item in a plurality of terms, and the accompanying information. Link information of at least one of internal link destination information in the knowledge database system and external link destination information outside the collective intelligence database system.
An item information processing apparatus that performs item integration processing, which is the first aspect,
(B) for two selected items in the collective intelligence database system, an attribute extracting unit that extracts the term and the link destination information as item attributes from information related to the two items;
(C) using the attribute extracted by the attribute extraction unit, an attribute calculation unit that calculates the similarity between the two items;
(D) an edit processing unit that determines whether or not the two items need to be integrated based on the similarity;
(E) a first communication unit for notifying a user or a system administrator of content that prompts the integration of the two items according to a determination result of the editing processing unit.

Moreover, the item information processing apparatus which is a 2nd aspect is
(F) An attribute extraction unit that extracts all the items registered in the collective intelligence database system as items to be processed for each item, and extracts the term and the link destination information from the items related information as item attributes When,
(G) For each of all combinations of two items created from all the items, using the attribute of the item used for the combination, an attribute calculation unit that acquires the similarity between the items;
(H) When two or more combinations of items whose acquired similarity is higher than a predetermined threshold value and the combination of each other shares a common item, the item used for the two or more combinations is 1 An edit processing unit that determines that integration into one item is required,
(I) a first communication unit for notifying a user or a system administrator of contents for prompting integration of items according to a determination result of the editing processing unit.

On the other hand, the item information processing apparatus that performs the item division processing according to the third aspect is
(J) a paragraph dividing unit that divides a sentence related to a selected item in the collective intelligence database system into a plurality of paragraphs;
(K) for each of the plurality of divided paragraphs, an attribute extraction unit that extracts the term and the link destination information as paragraph attributes;
(L) An attribute calculation unit that calculates the degree of similarity between paragraphs for all combinations of two paragraphs created from a plurality of paragraphs divided by the paragraph division unit, using the attributes of the paragraphs used for the combination When,
(M) an edit processing unit that determines whether or not the selected item needs to be divided based on the calculation result of the similarity;
(N) a first communication unit for notifying a user or a system administrator of content that prompts the user to divide the selected item according to the determination result of the editing processing unit.

  The item information processing apparatus and the item information processing method described above can accurately determine whether or not an item editing (integration or division) process is necessary and notify the user. Therefore, a collective intelligence database system that is easy to handle for users can be realized, and can be used by more users.

It is a figure which shows the structure of the outline of one Embodiment of the item information processing system using an item information processing apparatus. It is a figure explaining the outline | summary of the integration process performed in the item information processing system shown in FIG. It is a figure explaining the outline | summary of the division | segmentation process performed in the item information processing system shown in FIG. (A) And (b) is a figure which shows the example of the extraction result of the link address information and the extraction result of a word which the item information processing apparatus shown in FIG. 1 performs. It is a flowchart explaining the flow of the integration process shown in FIG. (A) And (b) is a figure which shows the example of the item used as integration object in the integration process shown in FIG. 5, and the example of the created integrated item plan. It is a figure which shows the other example of the integrated item plan different from FIG.6 (b). It is a flowchart explaining the flow of the division | segmentation process shown in FIG. (A) And (b) is a figure which shows the example of the item used as the division | segmentation object in the division | segmentation process shown in FIG. 8, and the example of the division item proposal produced.

  Hereinafter, the item information processing apparatus of the present invention will be described in detail based on embodiments. FIG. 1 is a schematic configuration diagram of an embodiment of an item information processing system using an item information processing apparatus.

An item information processing system 10 shown in FIG. 1 is a system that determines whether or not to edit an item registered in a collective intelligence database system such as “Wikipedia” and information related to the item. The collective intelligence database system includes a collective intelligence database server 14 and a user terminal 16. On the other hand, the item information processing system 12 includes an item information processing device 12 and a system administrator terminal 18 in addition to the collective intelligence database system.
Here, editing includes integration and division of items. In accordance with integration and division of items, integration and division of information regarding items is also performed. The information regarding each of the plurality of items registered in the collective intelligence database system includes a sentence describing the contents of each item in a plurality of terms and link destination information described accompanying the sentence. The link destination information includes at least one of internal link destination information in the collective intelligence database system and external link destination information outside the collective intelligence database system.
The determination of whether or not to integrate and divide items is performed by calculating the similarity. This similarity is calculated using the attribute included in the information regarding the item. An attribute characterizes an item, and includes terms included in information about the item, internal link address information, and external link address information.

(Outline of integrated processing of item information processing system)
FIG. 2 is a diagram for explaining integration processing among editing processing performed in the item information processing system 10. The integration process mainly includes determination of necessity of integration performed in the item information processing apparatus and creation of an integrated item plan. The integration process shown in FIG. 2 is an outline of the integration process when items A and B are registered and provided in the collective intelligence database system even though the items are actually the same “object”. Is explained.
First, the item information processing apparatus 12 of the item information processing system 10 selects information on the item A from the collective intelligence database system, and extracts external link address information, internal link address information, and terms. On the other hand, for item B, information related to item B is extracted from the collective intelligence database system, and external link address information, internal link address information, and terms are extracted. The item information processing apparatus 12 calculates similarity using the extracted external link address information, internal link address information, and terms. When the result of the similarity exceeds a predetermined threshold, the item information processing apparatus 12 creates an integrated item plan and notifies the user or system administrator of the prompt for the integration of item A and item B. When the item information processing apparatus 12 receives a reply of the integrated approval from the user or the system administrator, and further receives the approval of the integrated item proposal, the created integrated item proposal is fixed, and this integrated item proposal is the collective intelligence database server. 14 is registered as a new item.

(Outline of division processing of item information processing system)
FIG. 3 is a diagram for explaining the dividing process. The division processing mainly includes determination of necessity of division performed in the item information processing apparatus 12 and creation of a division item proposal. The division process shown in FIG. 3 is an outline of the division process when information about a plurality of “objects” is included in one item A even though the plurality of “objects” are different from each other. .
First, the item information processing apparatus 12 of the item information processing system 10 selects information on the item A from the collective intelligence database system, and divides this information into paragraph units. The paragraph of item A is assumed to include paragraphs 1-3. In this case, the item information processing apparatus 12 extracts external link address information, internal link address information, and terms in paragraphs 1 to 3, respectively. Using the extracted external link address information, internal link address information, and terms, the item information processing apparatus 12 calculates the similarity between paragraphs 1 and 2, paragraphs 2 and 3, and paragraphs 3 and 1. Further, the cluster analysis of the paragraph is performed based on the calculation result of the similarity. In the cluster analysis, the item information processing apparatus 12 obtains the number of clusters corresponding to the number of paragraphs to be divided based on the similarity calculation result, and determines that the item needs to be divided when the number of clusters is larger than one. At this time, the item information processing apparatus 12 creates a division item plan and notifies the user or system administrator of the division prompt. When a divisional approval is received from the user or the system administrator and the proposed divided item is further approved, the divided item plan is finalized and registered as a new item in the collective intelligence database server 14.

(Configuration of item information processing system)
As illustrated in FIG. 1, the item information processing system 10 includes an item information processing device 12, a collective intelligence database server 14, a plurality of user terminals 16, and a system administrator terminal 18.
The item information processing apparatus 12 determines whether it is necessary to edit items registered in the collective intelligence database server 14 and creates an integrated item plan and a divided item plan. The item information processing device 12 is connected to the collective intelligence database server 14 and the user terminal 16 via the network 20, and further connected to the system administrator terminal 18.

  The item information processing apparatus 12 is a computer including a CPU (arithmetic processing unit) 22, a memory 24, and a communication unit 26. When the program stored in the memory 24 is started, the following parts are formed as modules. That is, the item information processing apparatus 12 in which the program is activated has the attribute extraction unit 28, the attribute calculation unit 30, the edit processing unit 32, and the paragraph division unit 34 as software modules.

  For the plurality of items registered in the collective intelligence database server 14, the attribute extraction unit 28 selects a plurality of terms in the document, external link address information, and internal address link information from information related to each item. Extract as an attribute. The extracted attributes are recorded in the memory 24.

  FIG. 4A shows an example in which link destination address information is extracted from information on items. The information related to the item shown in FIG. 4A (in the upper frame) is an example of data shown in “Wikipedia” having “CP violation” as an item. Since the external link address information and the internal link address information are described in a predetermined format, the external link address information and the internal link address information can be easily extracted from this data. The extracted result is shown in the lower frame of FIG. These link address information is used to calculate the degree of similarity, which will be described later, as representing the attribute of the item.

  For a plurality of terms extracted from the text in the information related to each item, the text is first decomposed in units of words. FIG. 4B shows an example in which a sentence in the information related to the item is broken down into words. This processing is performed using a morphological analysis program (for example, “tea bowl”). For each of these words extracted by decomposition, a term group set in advance as an important term is searched. If there is a matching word as a result of the search, this word is used as an attribute of the item. . This term group includes, for example, proper nouns, technical terms divided into genres, common nouns, and the like. Therefore, in this case, terms used as attributes include proper nouns, technical terms, and common nouns.

The attribute calculation unit 30 calculates the similarity between two items using the attributes of each item extracted by the attribute extraction unit 28.
A known method is used to calculate the similarity. For example, “Similarity calculation using the vector space method”, Shogo Shibuya, Tomoyuki Ayasu, Mitsunori Miki,
(Http://nlp.nagaokaut.ac.jp/%E9%A1%9E%E4%BC%BC%E5%BA%A6: Searched on May 28, 2009) is used. The attribute extracted by the attribute extraction unit 28 can be represented by a vector to calculate the similarity.
The degree of similarity is represented by one real number of 0 or more and 1 or less which is obtained by inputting two items A and B (or a paragraph in the information about each item). When a function for obtaining one real number from two items (or paragraphs) A and B is represented as Similar (A, B), the Similar (,) has the following (Condition 1) to (Condition 3).
(Condition 1): For arbitrary items (or paragraphs) A and B, the value of Similar (A, B) takes a real value in the range of 1 at maximum and 0 at minimum.
(Condition 2): For any item (or paragraph) A, B, the value is the same even if the order of input of item (or paragraph) A, B is changed. Similar (A, B) = Similar (B, A)
(Condition 3): When the same item (or paragraph) A is given as two inputs to any item (or paragraph) A, B, the value is always 1 (but conversely the value is 1) But the two inputs are not necessarily the same).

(Similarity calculation method 1)
For such definition of similarity, for example, calculation of similarity when external link address information or internal link address information is used as an attribute is performed as follows.
Of the two items (or paragraphs) for which similarity is to be calculated, only two types of external link address information and internal link address information are extracted from item _Di (i-th item), and external link address information and internal link are extracted. A set {ν _i } having address information as an element is created. Similarly, a set {ν _j } is created from the other item (or paragraph) D _j (jth item). Here, the set {ν _i } and the set {ν _j } are sets in which only two types of external link address information and internal link address information of the items (or paragraphs) D _i and D _j are extracted and listed as a list. is there. Further, a set {ν _u } is obtained by merging the set {ν _i } and the set {ν _j } to remove duplicate elements.
Here, the following set {φ _i } is introduced as a “feature vector” characterizing each item (or each paragraph) D _i . That is, the number of items D _i that match the elements of the set {ν _u } is counted, and the count result is set as a set {φ _i }. Here, φ _ik (the k-th element of the set {φ _i }) = “the number that matches the k-th element of the set {ν _u } included in the item D _i ”. By this method, the feature vectors {φ _i } and {φ _j } of the items D _i and D _j are obtained. Thereby, the amount “Sim1” determined by the following equation (1) is set as the similarity between the item D _i and the item D _j .

That is, Expression (1) indicates that “the more the number of matching links described in two items is larger, the higher the similarity is”. Of course, “Sim1” satisfies the conditions 1 to 3 to be satisfied by the above-mentioned Similar function.
As described above, the similarity is calculated using “the number of elements included in the item D _i that matches the k-th element of the set {ν _u }”. It is an integer value when the link address information and the internal link address information completely match. However, in addition to this, as described below, when a part of the external link address information and the internal link address information match, a value of 0 to 1 may be given. For example, in the case of similarity between external link address information, the value 1 is different if the addresses are completely matched, and the value 0 is different if only the file name portion in the external link address information is different, and the other portions are matched. .7 may be given a value of 0.3 if only the server part in the external link address information matches.
In the case of the similarity between the internal ring address information, for example, the value 1 if the same item is linked, and the value if the genre of the link destination is defined for the internal ring address information and this genre matches. 0.8 may be given. In the above formula (1), | φ _i |, | φ _j | of the denominator is the total number of elements of {φ _i }, {φ _j }.

(Similarity calculation method 2)
Similarity calculation example 2 is an example of calculating similarity using terms, external link address information, and internal link address information.
A set of words, external link address information, and internal link address information that match terms listed in a predetermined term group list T are extracted from the item D _i out of the two items whose similarity is to be calculated { δ _i } is created. Similarly, a set {δ _j } is created from another item D _j . Here, the set {δ _i } and the set {δ _j } are a list obtained by extracting terms, external link address information, and internal link address information.
Similarity between the item D _i and item D _j uses the feature vectors [psi _i calculated from the item D _i, the following equation using the feature vectors [psi _j calculated from item D _j (2) Is calculated. Here, the value of the k-th element ψ _ik of the feature vector ψ _i is the same as the m-th element of the set {δ _i } and the set T ′ (T ′ is a set created in the same manner as in Example 1. The degree of similarity between {ν _u } and the k-th element constituting the term group {T} defined as a valid term in advance is calculated, and the calculation result is expressed as m It is a value obtained by summing while changing from 0 to the number of elements of the set {δ _i }. The feature vector ψ _j is obtained in the same manner.
In order to efficiently obtain such feature vectors ψ _i , ψ _j , it is preferable to calculate in advance the similarity between elements constituting the union T ′. That is, the similarity value between the i-th element T ′ _i and the j-th element T ′ _j may be calculated in advance.

Here, the value of the similarity Sim (,) between the elements of the above formula (2) is determined as follows, for example.
In the case of the similarity between the external link address information, for example, if the external link address information completely matches, the value 1 is different, only the file name portion is different, and if the other portions match, the value 0. 7 is assigned a value of 0.3 if only the server part matches.
In the case of similarity between internal ring address information, for example, if the same item is linked, the value is 1, and if the genre of the link destination is defined in the internal ring address information and this genre matches, the value is 0 .8 is given.
Furthermore, in the case of similarity between terms, a value of 1 is assigned if the terms are completely matched, and a value of 0.8 is assigned if the genre is predefined for the terms and the genre is matched.
・ In other cases, 0 is assigned.
As described above, the attribute calculation unit 30 calculates the similarity. The calculated similarity is recorded in the memory 24.

  The edit processing unit 32 determines the necessity of item integration and item division based on the similarity calculated by the attribute calculation unit 30. Further, when it is determined that the items need to be integrated in the integration processing, the editing processing unit 32 creates an integrated item plan in which two items are integrated into one item. That is, when the calculated similarity exceeds a predetermined threshold, an integrated item plan is created. In the creation of the integrated item plan, one of the two items is a main item, the other item is a subordinate item, and information about the subordinate item is merged with information about the main item. Of course, information related to the subordinate items may be included in the information related to the main item. The created integrated item proposal is notified to the user or the system administrator. In response to this notification, when the user and the system administrator are waiting for an integration response and an integration item plan response, the integration item plan is disclosed to the user as being temporarily integrated.

When the integration between the two items is confirmed, the editing processing unit 32 gives the integration processing target information to the new integrated item. The integration processing target information is information indicating that it is a target item for integration processing to be performed in the future. That is, the integration processing target information is used as a criterion for selecting an item when the attribute of the item is extracted and the similarity is calculated in the future. When the integration between the two items is confirmed, the editing processing unit 32 also assigns division processing target information indicating that the item is a target item of the division processing to be performed in the future to the new integrated item. This is because the integrated items may be divided in the future.
It should be noted that the integrated processing target information given to the items is calculated by calculating similarity between all items registered in the collective intelligence database server 14, and as a result, the calculated similarity exceeds any threshold. When there is not (when it is determined that integration is not required), the integration processing target information is removed from this item. Therefore, this item will no longer be subject to integration processing.

On the other hand, in the case of division processing, the edit processing unit 32 determines whether or not to divide items based on the result of similarity after calculating the similarity. Specifically, cluster analysis is performed based on the calculated similarity. In the cluster analysis, the editing processing unit 32 checks how many clusters a paragraph is divided into and determines whether or not to divide items according to the number of clusters. A known method is used for cluster analysis. For example, “Basics of Document Clustering”, Yoshiyo Onishi, Tomoyuki Kasuyasu, Mitsunori Miki,
The method proposed in (http://mikilab.doshisha.ac.jp/dia/research/report/2007/0913/004/report20070913004.html: searched on May 28, 2009) is used. Alternatively, cluster analysis using a known self-organizing map is used. Cluster analysis using self-organizing maps is, for example, “Basics of Self-Organizing Maps”, Toko Ito, Tomoyuki Suyasu, Mitsunori Miki,
(Http://mikilab.doshisha.ac.jp/dia/research/report/2005/0822/007/report20050822007.html: May 28, 2009 search) describes the method.

When it is determined that the item needs to be divided in the dividing process, the editing processing unit 32 creates a divided item plan in which the item is divided. That is, as a result of cluster analysis, if the number of clusters exceeds 1, a divided item plan is created. In creating a split item proposal, a new item is newly established and split. Such a division item proposal is notified to the user or the system administrator. When the divided item proposal is notified to the user or the system administrator, and the user and the system administrator are waiting for a division response and a divided item proposal response, the divided item proposal is disclosed to the user as being temporarily divided.
Further, when the division of the item is confirmed, the editing processing unit 32 assigns integrated processing target information indicating that the item is a target item of integration processing to be performed in the future to the new divided item. This is because the divided items may be integrated in the future. At this time, the division process target information is not given. The divided items are not further divided.

In the case of item division processing, the paragraph division unit 34 divides each paragraph of the text in the information related to the item separately. In the case of item integration processing, the paragraph division unit 34 does not function.
The paragraph division unit 34 performs division processing of a paragraph by selecting an item to which division processing target information has been previously assigned and which currently has the division processing target information. The division processing target information is used as a reference for selecting an item for performing the division processing.

In the case of integration processing, the communication unit 26 notifies the user terminal 16 or the system administrator terminal 18 to promote integration of two items when the calculated similarity is higher than a preset threshold. In the case of the division process, the communication unit 26 notifies the user terminal 16 or the system administrator terminal 18 so as to prompt the division of items when the number of clusters is larger than one.
When the communication unit 26 prompts the integration of items, the communication unit 26 simultaneously sends the integrated item plan created by the editing processing unit 32 to the user terminal 16 and the system administrator 18. In this case, when the user or the system administrator receives the notification of the prompt for integration, the user or the system administrator can return to the item information processing apparatus 12 whether the integration is approved and whether the integrated item proposal is approved. The edit processing unit 32 finalizes the integrated item plan according to the response received by the communication unit 26. Alternatively, when the communication unit 26 receives a reply that the integration is approved but the integration item proposal is not accepted, and receives the alternative proposed by the user or the system administrator, the editing processing unit 32 receives the received alternative Confirm as an integrated item. Alternatively, when the communication unit 26 receives a reply that the item integration is not approved, the integrated item proposal is deleted, and the original item and information about the item are maintained.

  In addition, when the communication unit 26 prompts the division of the item, the communication unit 26 simultaneously transmits the divided item plan created by the editing processing unit 32 to the user terminal 16 and the system administrator 18. In this case, when the user or the system administrator receives notification of the prompt for division, the user or the system administrator can reply to the item information processing apparatus 12 whether the division is approved or not and whether the divided item proposal is approved. The edit processing unit 32 finalizes the division item plan according to the response received by the communication unit 26. Alternatively, when the communication unit 26 receives a reply that the division is approved but the division item proposal is not accepted, and receives the alternative proposed by the user or the system administrator, the editing processing unit 32 receives the received alternative. Is confirmed as a split item. Alternatively, when the communication unit 26 receives a reply that it does not approve the division of the item, the divided item plan is deleted, and the original item and information regarding this item are maintained.

The collective intelligence database server 14 is a computer including a CPU (arithmetic processing unit) 42, a memory 44, a communication unit 46, and a database 48, and functions as a collective intelligence database system when a program stored in the memory 44 is activated. .
The collective intelligence database server 14 searches the items registered in the database 48 by being accessed from the user terminal 16 and receiving an item search instruction. Further, upon receiving access from the user terminal 16 or the system administrator terminal 18, items registered in the database 48 and information related to the items are edited (integrated or divided).

The user terminal 16 transmits information related to the item to the collective intelligence database server 14, further receives a notification of prompting for item integration or item division, and a communication unit that sends a response to the notification to the item information processing device 12. Have.
The system manager terminal 18 also has a communication unit that receives a notification of item integration or division prompting and sends a response to the notification to the item information processing apparatus 12.

(Item integration processing method)
FIG. 5 is a diagram illustrating a flow of item integration processing performed by the item information processing system 10. The integration process described below is performed intermittently or periodically.
First, the attribute extraction unit 28 of the item information processing device 12 selects one item to which the integration process target information is assigned from the database 48 of the collective intelligence database server 14 (step S10). This item is item a. The integration processing target information is information given to this item when a new item is newly established or when a new item is set by integration or division of items described later.
Next, the attribute extraction unit 28 selects an item other than the item a (referred to as item b) (step S20). The item b is selected from the items registered in the database 48 regardless of whether or not the integration processing target information is given.

  Next, the attribute extraction unit 28 extracts the attributes of item a and item b (step S30). As described above, the attribute includes a term, external link address information, and internal link address information. In terms of term extraction, first, a known morpheme program is used to decompose information (sentences) related to items into words, and the plurality of decomposed words are converted into terms in terms set in advance in the item information processing apparatus 12. The matching word is searched, and the matching word is extracted as a term that is an attribute of the item. On the other hand, the external link address information and the internal link address information are described in a predetermined format in the information relating to the item, so that all external link address information and internal link address information are extracted by searching the predetermined format. can do. Each term in the preset term group is defined as a word that characterizes the explanatory text of the item, and each genre is defined as a genre to which each term belongs. This genre is used when calculating the similarity as described above.

  Next, the attribute calculation unit 30 calculates the similarity between the item a and the item b using the extracted attribute (step S40). The similarity is calculated using the Sim1 or Sim2 function defined by the above formula (1) or (2). The calculated similarity result is stored in the memory 24.

Next, the edit processing unit 32 determines whether or not the calculated similarity is greater than a preset threshold value (step S50). If the degree of similarity is greater than the threshold value, the editing processing unit 32 determines that integration is required, and creates an item c obtained by integrating item a and item b as an integrated item plan (step S60).
When integrating the items, the integrated item plan is automatically created by merging the information related to the other subordinate item after the information related to the main item. The setting of the main item and subordinate items is obtained by weighting and adding the number of external link address information, the number of internal link address information, and the number of terms treated as attributes for each item. The item with the larger is the main item.
In the item c, for example, a marker is added to the portion where the information of the subordinate items is merged and displayed to the user or the like. Moreover, when the text in the information related to the main item is composed of a plurality of paragraphs, information related to the items that follow each paragraph may be arranged. Moreover, you may change the arrangement place for every paragraph.
If it is determined in step S50 that the similarity is equal to or less than the threshold value, the process proceeds to step S80 described later.

Next, the communication unit 26 sends a notification for urging the user terminal 16 to integrate the item a and the item b (step S70). This notification may be sent to the system administrator terminal 18. Along with this notification, item c is sent as an integrated item proposal. As will be described later, the user receives notification of prompting for integration and determines whether or not to approve integration. Approval approval is transmitted from the user terminal 16 as a response to the notification.
Next, for item b, it is determined whether all items other than item a registered in database 48 have been selected (step S80). If the result of the determination is negative, the item selected as the item b is changed, the new item is changed to the item b (step S90), and the process returns to step S30. If the result of the determination is affirmative, the selection of item b ends. Thereby, the integration process of item a ends. At this time, the integrated processing target information is deleted from the item a to which the integrated processing target information is assigned. When the integration process is performed in the future, this item a is not selected in step S10.

  Next, it is determined whether all items registered in the database 24 and to which the integration process target information is assigned have been selected (step S100). If the determination result is affirmative, the integration process ends. If the determination result is negative, the item selected as the item a is changed, a new item is selected as the item a (step S110), and the process returns to step S20.

If the above steps S10 to S100 are repeated and the determination in step S50 is affirmative, the communication unit 26 notifies the user terminal 16 of item integration (step S70). Receive from terminal 16. In this response, the edit processing unit 32 determines whether the user has approved the integration (step S120). If the user approves the integration of items and the integrated item proposal (item c), the integration is confirmed (step S130), and the integrated item proposal is registered in the database 48. If the user approves the integration but does not approve the integrated item proposal, the editing processing unit 32 determines the alternative returned from the user as an integrated item and registers it in the database 48. Items a and b that are targeted for integration are deleted from registration. However, the item a and the item b are not deleted from the database 48. This is because the items a and b registered in the past can be restored as needed in the future. Thereafter, the edit processing unit 32 gives the integration process target information and the division process target information to the integrated items (step S140). The reason why the integration processing target information and the division processing target information are given to the items is to allow the item information processing apparatus 12 to easily select the processing target items when performing the integration processing and the division processing in the future.
On the other hand, when the user does not approve the integration, a restoration process is performed (step S150). In the restoration process, the integrated item proposal is deleted, and the registration of the items a and b that are the targets of integration is maintained as they are.
The series of integration processes described above are performed periodically or intermittently.

  FIG. 6A shows an item “Rapa Nui”, an item “Easter Island”, and information on these items as examples registered in the database 48. Since “Rapa Nui” and “Easter Island” are actually the same “object”, it is inherently inappropriate to set two items. Actually, the similarity between the item “Rapa Nui” and the item “Easter Island” is larger than a preset threshold value, and it is determined that the items need to be integrated. Alternatively, the system administrator terminal 18 is notified.

FIG. 6B shows an example of an integrated item plan in which the item “Rapa Nui” and the item “Easter Island” are integrated. In this example, the item “Rapa Nui” is the main item, the item “Easter Island” is the subordinate item, and the information about the item “Easter Island” is followed by the information about the item “Rapa Nui”. Are connected. Such an integrated item plan is sent to the user terminal 16 or the system administrator 18. Also, in the first part of the information, “This description is a provisional integration of the item“ Rapa Nui ”and the item“ Easter Island ”.... Click“ Approve ”. "Is written. Such an integrated item proposal is disclosed to the user as being temporarily integrated.
FIG. 7 shows an example in which the integration of the item “Rapa Nui” and the item “Easter Island” is fused, not combined. Such a merged sentence may be temporarily edited by a user or system administrator and created as an integrated item proposal.

The above integration processing method is realized by calling a predetermined program from the memory 24 of the item information processing apparatus (computer) 12 and executing it.
In other words, the predetermined program is obtained by the CPU (arithmetic processing unit) 22 of the item arithmetic processing unit 12 regarding the two items selected from the collective intelligence database server 14 from the information on each item, the term, the link destination information, Are extracted as attributes and stored in the memory 24, the CPU 22 calculates the similarity between the two items using the extracted item attributes, and the CPU 22 calculates the similarity. And a signal for notifying the user or the system administrator that the CPU 22 prompts the integration of information regarding the two items according to the determination result of the necessity of item integration. The program which recorded the procedure to produce | generate.

As described above, in the item integration process, link address information is used in addition to the term. Therefore, it is necessary to integrate items compared to the case of calculating the similarity using only the term without using the link address information. Whether or not can be determined accurately.
In addition, the item information processing apparatus 12 creates an integrated item plan in which two items are integrated into one item according to the calculated similarity, thereby reducing the burden on the user to create the integrated item plan, Promote participation in knowledge database systems.
When the item information processing device 12 determines that the integration between the two items is confirmed, the item information processing device 12 gives the integration processing target information to the new integrated item, so that the item to be selected in the future integration processing can be easily identified. it can.
Further, when the item information processing apparatus 12 determines that the integration of the items is not required, the integration processing target information is deleted from the selected item, so that the burden of calculating similarity in the future integration processing is reduced. Can do.
When the item information processing device 12 determines the integration of two items, the item information processing device 12 assigns the division processing target information to the new integrated item, so that the item to be selected in the future division processing can be easily identified. .

(Modification)
The following is mentioned as a modification of the item information processing system 10 shown in FIG.
In the item information processing system 10, when performing the integration process, the attribute extraction unit 32 of the item information processing apparatus 12 extracts attributes for each item using all items registered in the collective intelligence database server 14 as processing target items. The attribute calculation unit 30 acquires similarities for all combinations of two items obtained from all items. The editing processing unit 32 is an item used for two or more combinations when the acquired similarity is two or more combinations of items higher than a predetermined threshold and the combination of each other shares an item common to each other It is determined that it is necessary to integrate the items into one item. The communication unit 26 notifies the user or system administrator of the content that prompts the integration of items according to the determination result of the editing processing unit 32. That is, when the degree of similarity exceeds a predetermined threshold and the items that are common to each other are shared, these items are integrated as one item.
As described above, when three or more items are integrated, the items can be efficiently integrated.

(Item split processing method)
FIG. 8 is a diagram illustrating a flow of item division processing performed by the item information processing system 10. The division processing described below is performed intermittently or periodically like the integration processing described above.

First, the attribute extraction unit 28 of the item information processing apparatus 12 selects one item to which the division process target information is assigned from the database 48 of the collective intelligence database server 14 (step S210). This item is item A. The division process target information is information indicating that the item is a target of the division process when a new item is newly established or when a new item is set by integrating the items described above.
Next, the paragraph dividing unit 34 divides the sentence in the information related to the selected item A into each paragraph (step S220). Since the paragraph is a unit obtained by dividing the sentence described in the information related to the item A by a line feed, the paragraph can be easily divided for each paragraph.

Next, the attribute extraction unit 28 extracts the attribute of each paragraph for each divided paragraph, and the attribute calculation unit 30 calculates the similarity between the paragraphs (step S230). The attribute extraction is performed in the same manner as the item attribute extraction in the integration process. Regarding the extraction of terms, information (sentences) related to items is first decomposed into words using a known morpheme program, and the words that match the terms in the term group set in advance in the item information processing apparatus 12 for the decomposed words And matching words are extracted as item attributes. On the other hand, the external link address information and the internal link address information are described in a predetermined format in the item-related information. By searching the format, all external link address information and internal link address information are attributed. Can be extracted.
In calculating the similarity between paragraphs, the similarity is calculated using the above-described equations (1) and (2) using the attributes of each paragraph, similarly to the calculation of the similarity using the item attributes. . The similarity calculation result is stored in the memory 24.

Next, the edit processing unit 32 performs a cluster analysis on the paragraph (step S240). In the cluster analysis, paragraphs are divided into clusters using the calculated result of similarity between the paragraphs. The cluster analysis method is performed by the method described above. By cluster analysis, it is divided into one or a plurality of clusters.
Next, the edit processing unit 32 determines whether the number of clusters is greater than 1, that is, whether the number of clusters is 2 or more or 1 (step S250). The number of clusters of 2 or more means that there are 2 or more clusters. Therefore, the paragraph is divided in cluster units. If the number of clusters is 2, the item is divided into two, and if the number of clusters is 3 or more, the item is divided into 3 or more.

If the determination in step S250 is affirmative, that is, if the number of clusters is two or more, the editing processing unit 32 newly establishes an item according to the number of clusters and quotes information on paragraphs belonging to the cluster. Information about the item (item C) is created (step S260). In this way, a division item plan is created. On the other hand, if the determination in step S250 is negative, that is, if it is determined that item division is not necessary, the process proceeds to step S280 described later.
Next, the communication unit 26 notifies the user terminal 16 of the prompt for dividing item A (step S270). At this time, the division item plan created in step S260 is transmitted at the same time to request division and approval of the division item plan.

  Next, the edit processing unit 32 determines whether or not all items to which the division process target information registered in the database 48 has been selected for the item A (step S280). If all items have not been selected, the item selected as item A is changed to another item and selected as item A (step S290), and the process returns to step S220.

When the above steps S210 to S290 are repeated and the determination in step S250 is affirmative, the communication unit 26 notifies the user terminal 16 to divide items (step S270). Receive from terminal 16. Based on this response, the editing processing unit 32 determines whether or not the user has approved the division (step S300).
When the user approves the division and the division item proposal, the division is confirmed (step S310), and the division item proposal newly established according to the cluster is registered in the database 48. If the user approves the division but does not approve the divided item proposal, the editing processing unit 32 determines the alternative returned from the user as the divided item and registers it in the database 48. On the other hand, the text and link address information quoted in the split item proposal in the item A to be split are removed from the item A. Registration is maintained for item A from which some of the information in the item has been removed. Thereafter, the edit processing unit 32 gives the integration process target information to the newly created division item plan (step S320). The integrated processing information is assigned to the item A to be divided, and no division processing information is given. Since item A is the remaining item that has already been divided, no division processing target information is given. The reason why the integration process target information is given is to make the item A an item to be processed when the integration process is performed in the future.
On the other hand, if the user does not approve the division in step S300, a restoration process is performed (step S330). In the restoration process, the item C is deleted and the registration of the item A is maintained. At this time, the integration processing target information of item A is removed.

  FIG. 9A shows an item “virus” and information on these items as an example registered in the database 48. The item "virus" is a minute structure that can replicate itself using cells of other organisms, an "object" that represents an organism causing a disease such as influenza, and a computer virus that represents " Object ". That is, two different “objects” are included in one item. For this reason, by performing the above-described division processing, the number of clusters becomes 2, and the item information processing apparatus 12 notifies the item division prompting.

FIG. 9B shows an example in which a new item C is newly established as “virus (2)” as a proposed division item. In the item “virus (1)”, the paragraph regarding the computer virus is removed from the information regarding the item “virus”.
Such a division item plan is sent to the user terminal 16. In addition, in the first part of the information, “This description is the item“ virus ”separated into items“ virus (1) ”and“ virus (2) ”. . "Is written. Such a division item proposal is disclosed to the user as being temporarily divided. While confirming the divided item proposal, the user can reply whether or not the divided item proposal is appropriate.

The above division processing method is realized by calling and executing a predetermined program from the memory 24 of the item information processing apparatus (computer) 12.
In other words, a predetermined program is stored in the memory 24 by the CPU (arithmetic processing unit) 22 of the item information processing apparatus (computer) 12 dividing a sentence related to the selected item in the collective intelligence database server 14 into paragraphs. The procedure, the CPU 22 extracts the term and link destination information as attributes for each of the plurality of divided paragraphs and stores them in the memory 24, and the CPU 22 stores the two paragraphs created from the plurality of divided paragraphs. For each of all combinations, a procedure for calculating the similarity between paragraphs using the attributes of the paragraphs used for the combination, a procedure for the CPU 22 determining whether or not to divide items based on the calculated similarity, and CPU 22 However, the content that prompts the user to divide the item selected according to the determination result of whether or not the item needs to be divided A program recording and a step of generating a signal to notify the user.

As described above, in the item division process, link address information and terms are used, so the necessity of item division is more accurate than when the similarity is calculated using terms without using link address information. Can be determined.
In addition, the item information processing apparatus 12 performs cluster analysis of a plurality of paragraphs divided based on the calculated similarity, and determines whether or not the items need to be divided based on the result of the cluster analysis. Therefore, it is possible to objectively and accurately determine whether or not to divide.
Since the item information processing apparatus 12 selects the item to which the division processing target information is currently assigned as the item to be divided, the item can be easily selected from a plurality of items.
The item information processing device 12 creates a division item plan for the item to be divided according to the result of the cluster analysis, thus reducing the burden of creating the integrated item plan for the user and promoting participation in the collective intelligence database system To do.
When the item information processing device 12 determines the division of the item, the item information processing device 12 assigns the integration processing target information to the new divided item, so that the item to be selected in the future integration processing can be easily identified.

  Regarding the above embodiment, the following supplementary notes are disclosed.

(Appendix 1)
An item information processing apparatus that processes items provided from a plurality of users through a network and registered in a collective intelligence database system,
The information related to each of a plurality of items registered in the collective intelligence database system is described in a sentence describing the contents of each item in a plurality of terms and accompanying the sentences, the collective intelligence database system Link information of at least one of internal link destination information and external link destination information outside the collective intelligence database system,
An attribute extraction unit that extracts the term and the link destination information as attributes of the item from the information about the two items for the two selected items in the collective intelligence database system;
Using the attribute extracted by the attribute extraction unit, an attribute calculation unit that calculates the similarity between the two items;
An edit processing unit that determines whether or not the two items need to be integrated based on the similarity;
An item information processing apparatus, comprising: a first communication unit that notifies a user or a system administrator of content that prompts integration of the two items according to a determination result of the editing processing unit.

(Appendix 2)
The item information processing apparatus according to attachment 1, wherein the editing processing unit creates an integrated item plan in which the two items are integrated into one item according to the calculated similarity.

(Appendix 3)
One item of the two items selected by the attribute extraction unit is an item to which integration processing target information indicating that it is a target item of item integration processing is given,
When the integration between the two items is confirmed according to the response content of the integration approval from the user or system administrator in response to the notification, the editing processing unit adds the integration processing target information to the new integrated item. Item information processing apparatus according to appendix 1 or 2, which is given.

(Appendix 4)
The attribute calculation unit calculates the similarity between all items registered in the collective intelligence database system for the item to which the integration processing target information is assigned. Item information processing apparatus of Claim 3 which deletes the said integration process target information from the said selected item, when it determines with integration not being required.

(Appendix 5)
When the integration of the two items is determined according to the response content from the user or system administrator to the notification, the editing processing unit is an item to be subjected to the item dividing process into a new integrated item. Item information processing apparatus given in any 1 paragraph of appendices 1-4 which gives division processing object information which shows this.

(Appendix 6)
An item information processing apparatus for processing items provided through a network from a plurality of users and registered in a collective intelligence database system,
The information related to each of a plurality of items registered in the collective intelligence database system is described in a sentence describing the contents of each item in a plurality of terms and accompanying the sentences, the collective intelligence database system Link information of at least one of internal link destination information and external link destination information outside the collective intelligence database system,
All items registered in the collective intelligence database system are items to be processed, and for each item, an attribute extraction unit that extracts the term and the link destination information from the information about the item as item attributes;
For each of all combinations of two items created from all the items, using the attribute of the item used for the combination, an attribute calculation unit that acquires the similarity between the items,
When two or more combinations of items whose acquired similarities are higher than a predetermined threshold and the combinations of the items share a common item, the items used for the two or more combinations are combined into one item. An edit processing unit that determines that integration is required;
An item information processing apparatus comprising: a first communication unit for notifying a user or a system administrator of contents for prompting integration of items according to a determination result of the editing processing unit.

(Appendix 7)
An item information processing apparatus that processes items provided from a plurality of users through a network and registered in a collective intelligence database system,
The information related to each of the plurality of items registered in the collective intelligence database system includes a sentence describing the contents of each item in a plurality of terms, and is written along with the sentences. Link information of at least one of internal link destination information and external link destination information outside the collective intelligence database system,
A paragraph dividing unit that divides a sentence related to the selected item in the collective intelligence database system into a plurality of paragraphs;
For each of the plurality of divided paragraphs, an attribute extraction unit that extracts the term and the link destination information as paragraph attributes;
For each of all combinations of two paragraphs created from a plurality of paragraphs divided by the paragraph division unit, an attribute calculation unit that calculates the similarity between paragraphs using the attributes of the paragraphs used for the combination;
An editing processing unit that determines whether or not the selected item needs to be divided based on the calculation result of the similarity;
An item information processing apparatus comprising: a first communication unit that notifies a user or a system administrator of content that prompts the user to divide the selected item according to a determination result of the editing processing unit.

(Appendix 8)
The edit processing unit performs a cluster analysis of the plurality of divided paragraphs based on the calculated similarity, and determines whether or not the selected item needs to be divided based on a result of the cluster analysis. Item information processing apparatus described in 1.

(Appendix 9)
The item information processing apparatus according to appendix 7 or 8, wherein the item selected by the paragraph division unit is an item to which division processing target information indicating that it is a target item of item division processing is assigned.

(Appendix 10)
The item information processing apparatus according to any one of appendices 7 to 9, wherein the editing processing unit creates a division item plan for an item to be divided according to the result of the cluster analysis.

(Appendix 11)
When the division of the item is confirmed according to the response of the division approval from the user or the system administrator in response to the notification, the editing processing unit adds a new item to the target item of the item integration process. Item information processing apparatus given in any 1 paragraph of appendices 7-10 which give integration processing object information which shows that there is.

(Appendix 12)
An item information processing system comprising a collective intelligence database server, an item information processing apparatus, and a user terminal, which processes items provided through a network from a plurality of users and registered in the collective intelligence database system,
The collective intelligence database server
A second communication unit that communicates with the user terminal and the item information processing apparatus, and a database that registers the item,
The information about each of the plurality of items registered in the database includes a sentence describing the contents of each item in a plurality of terms, and an internal part of the collective intelligence database server that is written along with the sentence. Link destination information and link destination information of at least one of external link destination information outside the collective intelligence database server,
The item information processing apparatus includes:
An attribute extraction unit for extracting the term and the link destination information as attributes of the item from information on each item for the two selected items in the database;
Using the attribute of each item extracted by the attribute extraction unit, an attribute calculation unit that calculates the similarity between the two items;
An edit processing unit that determines whether or not the two items need to be integrated based on the similarity;
A first communication unit for notifying a user or a system administrator of the content that prompts the integration of the two items according to the determination result of the editing processing unit,
The user terminal includes a third communication unit that transmits information about an item to the collective intelligence database server, further receives the notification, and sends a response to the notification to the item information processing apparatus. Characteristic item information processing system.

(Appendix 13)
An item information processing system comprising a collective intelligence database server, an item information processing apparatus, and a user terminal, which processes items provided through a network from a plurality of users and registered in the collective intelligence database system,
The collective intelligence database server
A second communication unit that communicates with the user terminal and the item information processing apparatus, and a database that registers the item,
The information about each of the plurality of items registered in the database includes a sentence describing the contents of each item in a plurality of terms, and an internal part of the collective intelligence database server that is written along with the sentence. Link destination information and link destination information of at least one of external link destination information outside the collective intelligence database server,
The item information processing apparatus includes:
A paragraph divider for dividing a sentence related to the selected item in the database into a plurality of paragraphs;
For each of the plurality of divided paragraphs, an attribute extraction unit that extracts the term and the link destination information as paragraph attributes;
For each of all combinations of two paragraphs created from a plurality of paragraphs divided by the paragraph division unit, an attribute calculation unit that calculates the similarity between paragraphs using the attributes of the paragraphs used for the combination;
An editing processing unit that determines whether or not the selected item needs to be divided based on the calculation result of the similarity;
A first communication unit for notifying a user or a system administrator of content for prompting division of the selected item according to a determination result of the editing processing unit;
The user terminal includes a third communication unit that transmits information about an item to the collective intelligence database server, further receives the notification, and sends a response to the notification to the item information processing apparatus. Characteristic item information processing system.

(Appendix 14)
An item information processing method for processing items provided from a plurality of users through a network and registered in a collective intelligence database system,
The information related to each of the plurality of items registered in the collective intelligence database system includes a sentence describing the contents of each item in a plurality of terms, and is written along with the sentences. Link information of at least one of internal link destination information and external link destination information outside the collective intelligence database system,
For the two selected items in the collective intelligence database system, extracting the term and the link destination information from the information about each item as item attributes;
Calculating the degree of similarity between the two items using the attribute of the extracted item;
Determining whether to integrate the two items based on the similarity;
An item information processing method comprising: notifying a user or a system administrator of contents for prompting the integration of the two items according to the determination result of the necessity of the integration.

(Appendix 15)
An item information processing method for processing items provided from a plurality of users through a network and registered in a collective intelligence database system,
The information related to each of the plurality of items registered in the collective intelligence database system includes a sentence describing the contents of each item in a plurality of terms, and is written along with the sentences. Link information of at least one of internal link destination information and external link destination information outside the collective intelligence database system,
Dividing the sentence about the selected item in the collective intelligence database system into a plurality of paragraphs;
For each of the plurality of divided paragraphs, extracting the term and the link destination information as paragraph attributes;
For each of all combinations of two paragraphs created from a plurality of paragraphs divided by the paragraph division unit, calculating the similarity between paragraphs using the attributes of the paragraphs used for the combination;
Determining whether the item needs to be divided based on the calculated similarity;
An item information processing method comprising: notifying a user or a system administrator of content that prompts the user to divide the selected item according to the determination result of the necessity of division.

(Appendix 16)
When a computer processes items provided through a network from multiple users and registered in the collective intelligence database system,
The information related to each item registered in the collective intelligence database system includes a sentence describing the contents of each item in a plurality of terms, and an internal part of the collective intelligence database system that is written along with the sentence. Link destination information and link destination information of at least one of external link destination information outside the collective intelligence database system,
The computing processing unit of the computer extracts the term and the link destination information from the information about each item for the two selected items in the collective intelligence database system as item attributes, and the computer To store in the memory of
A procedure in which the arithmetic processing unit calculates a similarity between the two items using the extracted attribute;
A procedure in which the arithmetic processing unit determines necessity of integration of the two items based on the similarity.
A computer that causes the arithmetic processing unit to generate a signal for notifying a user or a system administrator of a content that prompts integration of the two items in accordance with a determination result of the necessity of integration; An executable program.

(Appendix 17)
When a computer processes items provided through a network from multiple users and registered in the collective intelligence database system,
The information related to each item registered in the collective intelligence database system includes a sentence describing the contents of each item in a plurality of terms, and an internal part of the collective intelligence database system that is written along with the sentence. Link destination information and link destination information of at least one of external link destination information outside the collective intelligence database system,
A procedure in which the arithmetic processing unit of the computer divides a sentence related to a selected item in the collective intelligence database system into a plurality of paragraphs and stores the same in a memory of the computer;
For each of the plurality of divided paragraphs, the term and the link destination information are extracted as paragraph attributes and stored in the memory of the computer;
The arithmetic processing unit calculates, for each of all combinations of two paragraphs created from a plurality of divided paragraphs, the similarity between paragraphs using the attribute of the paragraph used for the combination;
A procedure in which the arithmetic processing unit determines whether or not to split the selected item based on the calculated similarity;
Causing the computer to execute a procedure for generating a signal for notifying a user or a system administrator of the content that prompts the user to divide the selected item according to the determination result of whether or not the division is necessary, A computer executable program.

  The item information processing apparatus of the present invention has been described in detail above. However, the item information processing apparatus of the present invention is not limited to the above-described embodiment, and various improvements and modifications can be made without departing from the gist of the present invention. Of course it is also good.

10 Item Information Processing System 12 Item Information Processing Device 14 Collective Intelligence Database Server 16 User Terminal 18 System Manager Terminal 20 Network 22, 42 CPU
24, 44 Memory 26, 46 Communication unit 28 Attribute extraction unit 30 Attribute calculation unit 32 Edit processing unit 34 Paragraph division unit 48 Database

Claims (11)

An item information processing apparatus that processes items provided from a plurality of users through a network and registered in a collective intelligence database system,
The information related to each of a plurality of items registered in the collective intelligence database system is described in a sentence describing the contents of each item in a plurality of terms and accompanying the sentences, the collective intelligence database system Link information of at least one of internal link destination information and external link destination information outside the collective intelligence database system,
An attribute extraction unit that extracts the term and the link destination information as attributes of the item from the information about the two items for the two selected items in the collective intelligence database system;
Using the attribute extracted by the attribute extraction unit, an attribute calculation unit that calculates the similarity between the two items;
An edit processing unit that determines whether or not the two items need to be integrated based on the similarity;
An item information processing apparatus, comprising: a first communication unit that notifies a user or a system administrator of content that prompts integration of the two items according to a determination result of the editing processing unit.   The item information processing apparatus according to claim 1, wherein the edit processing unit creates an integrated item plan in which the two items are integrated into one item according to the calculated similarity. One item of the two items selected by the attribute extraction unit is an item to which integration processing target information indicating that it is a target item of item integration processing is given,
When the integration between the two items is confirmed according to the response content of the integration approval from the user or system administrator in response to the notification, the editing processing unit adds the integration processing target information to the new integrated item. The item information processing apparatus according to claim 1, which is given.   The attribute calculation unit calculates the similarity between all items registered in the collective intelligence database system for the item to which the integration processing target information is assigned. The item information processing apparatus according to claim 3, wherein when it is determined that no integration is required, the integration processing target information is deleted from the selected item.   When the integration of the two items is determined according to the response content from the user or system administrator to the notification, the editing processing unit is an item to be subjected to the item dividing process into a new integrated item. The item information processing apparatus according to any one of claims 1 to 4, wherein division processing target information indicating that is provided. An item information processing apparatus for processing items provided through a network from a plurality of users and registered in a collective intelligence database system,
The information related to each of a plurality of items registered in the collective intelligence database system is described in a sentence describing the contents of each item in a plurality of terms and accompanying the sentences, the collective intelligence database system Link information of at least one of internal link destination information and external link destination information outside the collective intelligence database system,
All items registered in the collective intelligence database system are items to be processed, and for each item, an attribute extraction unit that extracts the term and the link destination information from the information about the item as item attributes;
For each of all combinations of two items created from all the items, using the attribute of the item used for the combination, an attribute calculation unit that acquires the similarity between the items,
When two or more combinations of items whose acquired similarities are higher than a predetermined threshold and the combinations of the items share a common item, the items used for the two or more combinations are combined into one item. An edit processing unit that determines that integration is required;
An item information processing apparatus comprising: a first communication unit for notifying a user or a system administrator of contents for prompting integration of items according to a determination result of the editing processing unit. An item information processing apparatus that processes items provided from a plurality of users through a network and registered in a collective intelligence database system,
The information related to each of the plurality of items registered in the collective intelligence database system includes a sentence describing the contents of each item in a plurality of terms, and is written along with the sentences. Link information of at least one of internal link destination information and external link destination information outside the collective intelligence database system,
A paragraph dividing unit that divides a sentence related to the selected item in the collective intelligence database system into a plurality of paragraphs;
For each of the plurality of divided paragraphs, an attribute extraction unit that extracts the term and the link destination information as paragraph attributes;
For each of all combinations of two paragraphs created from a plurality of paragraphs divided by the paragraph division unit, an attribute calculation unit that calculates the similarity between paragraphs using the attributes of the paragraphs used for the combination;
An editing processing unit that determines whether or not the selected item needs to be divided based on the calculation result of the similarity;
An item information processing apparatus comprising: a first communication unit that notifies a user or a system administrator of content that prompts the user to divide the selected item according to a determination result of the editing processing unit. )
The editing processing unit performs cluster analysis of the plurality of divided paragraphs based on the calculated similarity, and determines whether or not the selected item needs to be divided based on a result of the cluster analysis. Item information processing apparatus according to 7.   The item information processing apparatus according to claim 7 or 8, wherein the item selected by the paragraph division unit is an item to which division processing target information indicating that the item is a target item of the item division processing is added.   The item information processing apparatus according to claim 7, wherein the editing processing unit creates a division item plan for an item to be divided according to a result of the cluster analysis.   When the division of the item is confirmed according to the response of the division approval from the user or the system administrator in response to the notification, the editing processing unit adds a new item to the target item of the item integration process. The item information processing apparatus according to any one of claims 7 to 10, to which integrated processing target information indicating that there is present is assigned.