JP2015125764A - Information collection program, information collection method, and information collection device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently select appropriate information from a plurality of information and use the selected information.SOLUTION: The present invention acquires a plurality of collection information each including at least one subject character string among a plurality of subject character strings that include meaning belonging to a prescribed subject field, as well as correct answer knowledge information which is external knowledge information including the explanation of meaning of the subject character string in the subject field and incorrect answer knowledge information which is external knowledge information including the explanation of meaning of the subject character string outside the subject field. The present invention determines a first rank of each of the plurality of collection information in accordance with the level of the degree of mutual association in the plurality of collection information and incorrect answer knowledge information. Further, the present invention determines a second rank of each of the plurality of collection information in accordance with the level of the degree of mutual association in the plurality of collection information and incorrect answer knowledge information. Then, the present invention adjusts the first rank on the basis of a difference between the first rank and the second rank and determines a final rank of each of the plurality of collection information.

Description

  The present invention relates to a technique for collecting information in a computer.

As one aspect of information collection, it is generally performed that information search is performed using a character string related to information to be collected and information including the character string is collected.
As an example of a conventional technique for improving the accuracy of information retrieval, there is the following technique. In this technique, designation of a correct answer or an incorrect answer for query related information related to a query is received from a user. Then, new queries based on the specified correct answer or incorrect answer data are obtained, respectively, and query related information corresponding to the new query is obtained. Furthermore, the same data appearing in each query related information is attached based on the rank of each data in each query related information obtained corresponding to a new query based on correct or incorrect data. Data for output is generated by integrating the ranks.

JP 2010-108477 A

Here, for example, when information of various fields is included in the information to be searched, if the search is simply performed with a specific character string, the following problem may occur. That is, when a character string to be searched has a plurality of different meanings, information other than information that should be originally used is also extracted as a search result. For this reason, if information extracted in this way is simply collected, especially when the number of collected information is large, information to be originally used cannot be efficiently selected, and work efficiency may be poor.
Therefore, an object of one aspect of the present invention is to make it possible to efficiently select and use appropriate information from a plurality of collected information.

  In one aspect of the present invention, among a plurality of target character strings including meanings belonging to a predetermined target field, a plurality of collection information each including at least one target character string, and the meaning of the target character string in the target field Correct knowledge information, which is external knowledge information including explanation, and incorrect knowledge information, which is external knowledge information including meaning explanations outside the target field of the target character string, are acquired. And according to the high degree of mutual relation in a plurality of collection information and correct answer knowledge information, the 1st order of each of a plurality of collection information is determined. Further, the second order of each of the plurality of collection information is determined in accordance with the degree of mutual relation between the plurality of collection information and the incorrect knowledge information. Then, the first rank is adjusted based on the difference between the first rank and the second rank, and the final rank of each of the plurality of collection information is determined.

  According to one aspect of the present invention, it is possible to efficiently select and use appropriate information from a plurality of pieces of information.

It is explanatory drawing which shows an example of the network of collection information. It is explanatory drawing which shows an example of the network which added external knowledge information (correct knowledge information) to collection information. It is explanatory drawing which shows an example of the network which added external knowledge information (incorrect answer knowledge information) to collection information. It is explanatory drawing which shows an example of the whole structure of the system in this embodiment, and a function structure of an information collection server. It is explanatory drawing which shows an example of collection information. It is explanatory drawing which shows an example of external knowledge information, (A) shows an example of correct answer knowledge information, (B) shows an example of incorrect answer knowledge information. It is explanatory drawing which shows an example of the table which stores the order | rank of collection information, (A) shows an example of a correct answer order table, (B) shows an example of an incorrect answer order table. It is explanatory drawing which shows an example of the adjustment order table which stores the order | rank of collection information. It is a flowchart which shows an example of the whole process performed with an information collection server. It is a flowchart which shows an example of the order adjustment process performed with an information collection server. It is explanatory drawing which shows the specific example regarding adjustment of the order | rank of collection information. It is explanatory drawing which shows the specific example regarding adjustment of the order | rank of collection information. It is explanatory drawing which shows an example of the whole structure of the system in this embodiment, and a function structure of an information collection server. It is explanatory drawing which shows an example of the original order table which stores the order | rank of collection information. It is a flowchart which shows an example of the whole process performed with an information collection server. It is a flowchart which shows an example of the order adjustment process performed with an information collection server. It is explanatory drawing which shows the specific example regarding adjustment of the order | rank of collection information. It is explanatory drawing which shows the specific example regarding adjustment of the order | rank of collection information. It is explanatory drawing which shows the specific example regarding adjustment of the order | rank of collection information. It is explanatory drawing which shows the specific example regarding adjustment of the order | rank of collection information. It is an example of the hardware constitutions of the information collection server in this embodiment.

[Background and Outline of this Embodiment]
In recent years, consumers and the like have transmitted various information through social media and the like. For this reason, for example, companies that sell products and the like have a demand for analyzing consumer needs from various information transmitted by consumers and the like. For this reason, information transmitted by consumers or the like is searched with a character string (for example, a word or a combination of one or a plurality of words) estimated to be included in the information to be analyzed, and the character string. Collecting and using related information including In the following description, an example in which a character string to be searched is a word will be described. However, the technique described in this embodiment can be applied even to a search using a character string other than a word.

Here, the social media transmitted by such consumers and the like includes information of various fields. For this reason, when a word to be searched has a plurality of different meanings, information other than information that should be collected may be extracted as a search result.
For example, it is assumed that the word is the name of a product and the name of the product is the same as the name of a sports team not related to the product. In this case, when information search is performed using the name of the product, not only information related to the product but also information related to the sports team is extracted. For this reason, there are cases where the information that is originally desired cannot be efficiently selected. In addition, such a problem is a problem that may occur not only in a case where information transmitted by a consumer or the like through social media or the like is a target but also in a case where information is collected and used.

  Therefore, for example, the following technique can be used in order to enable efficient selection of information used in the meaning of the purpose of using information as a character string to be searched. That is, in this technique, ranking is performed based on the degree of relevance between a plurality of pieces of information. Specifically, first, at least one target character string is included among a plurality of character strings (hereinafter referred to as a target character string) including meanings belonging to a field for which information is collected (hereinafter referred to as a target field). Collect multiple pieces of information. And about this some collection information, the collection information containing the same character string is linked | related and a network is produced | generated. The character string here includes character strings other than the target character string (for example, other words and verbs). At this time, the degree of association is strengthened between the collected information including the same character string. And about the said some collection information, ranking (ranking) according to the high degree of mutual relation between collection information is performed. In this ranking, collection information associated with other collection information with a high degree of relevance is ranked higher. For details of the above technology, see, for example, `` ChiWang, Kaushik Chakrabarti, Tao Cheng, and Surajit Chaudhuri. (2012) Targeted disambiguation of ad-hoc, homogeneour sets of named entities.In Proceeding of the 21st international conference on World Wide Web References such as US Pat. No. 6,285,999 and “Laurence Page, Sergey Brin, Rajeev Motwani, and Terry Winograd. (1999) The Pagerank citation ranking: Bringing Order to the Web. Technical Report. Stanford InfoLab.” It is described in.

  In the ranking in the above technique, when the target character string includes a meaning other than the target field, collected information other than the target field may also be included in the ranking target. However, according to the above technique, as the information includes more character strings in common with other collected information, the degree of association with the other collected information is higher and the rank is higher. In general, collected information belonging to the same field often includes the same character string. For this reason, in the ranking, collection information that includes more content that matches the target field (collection information that is estimated to have a high degree of matching to the target field) is ranked higher. If collected information is selected using this order, it is possible to efficiently use collected information having a meaning suitable for the purpose.

  FIG. 1 shows a specific example of a network of collected information generated by the above technique. In this specific example, it is assumed that the target field is “sweets”. In addition, target character strings including meanings belonging to the “sweets” are “◯◯” and “ΔΔ”. Here, “OO” includes the meaning of the name of the soccer team in addition to the meaning of sweets. Further, “ΔΔ” includes the meaning of the country name in addition to the meaning of sweets. Therefore, it is assumed that collection information A to collection information D are collected as collection information including at least one of “XX” or “ΔΔ”.

Among these collection information A to collection information D, as shown in FIG. 1, the collection information A and the collection information C are contents related to sweets. These two pieces of information include, in common, character strings such as “eat” and “taste” that are often used in the content related to sweets. Thus, since information belonging to the same field includes many common character strings, the degree of association tends to be relatively high and the ranking tends to be higher.
On the other hand, the collection information B is about soccer, and there are few character strings in common with other collection information. For this reason, the relevance is relatively low. The collected information D is also about the country, and similarly, the degree of association is relatively low.

Here, as one piece of information that is generally highly reliable, external knowledge information (for example, information provided by Wikipedia (registered trademark), etc.) including an explanation (definition explanation) of the meaning of a word (character string), etc. )
For this reason, in the above ranking, in order to rank information suitable for the target field in a higher rank, external knowledge information including a description of the meaning of the target character string in the target field is generated as a network between the information. It has been added to the subject. By doing so, information that contains many character strings in common with the external knowledge information, that is, information that is highly compatible with the target field has a high degree of association with the external knowledge information, and as a result, ranks higher. It is expected to be.

  FIG. 2 shows external knowledge information including explanations of meanings in the field of “sweets” of “XX” and “ΔΔ” which are words including meanings belonging to “sweets” with respect to the network of FIG. A specific example of a network to which certain correct knowledge information A and correct knowledge information B are added is shown. As shown in the specific example, for example, the external knowledge information A including an explanation of the meaning of “XX” has a character string such as “candy” or “taste” in common with the collected information A. For this reason, the collection information A has a relatively high degree of association with the correct knowledge information A, and as a result, the rank is expected to be higher. For example, the correct knowledge information B including an explanation of the meaning of “ΔΔ” has a character string such as “ABC Confectionery” or “Chocolate” in common with the collected information C. For this reason, the collection information C has a relatively high degree of relevance with the correct knowledge information B, and is similarly expected to be ranked higher. Note that not only the combination of these, but also the collected information C and the external knowledge information A, for example, are content related to the same confectionery field, so the word “taste” is common. In this way, by adding external knowledge information including semantic explanation in the target field to the network, the degree of relevance with other collected information in the collected information including the contents of the target field is relatively high, and the rank is higher. Become. On the other hand, for example, the collected information B, which is the content related to the soccer team, and the collected information D, which is the content related to the country, have few character strings in common with the correct answer knowledge information A and the correct answer knowledge information B, and have a relatively low degree of association.

  However, the external knowledge information including the explanation of the meaning in the target field of the target character string including the meaning belonging to the target field is utilized in this way, but the external knowledge including the meaning of “non-target field” is used. Information is not utilized. Note that the external knowledge information including the meaning explanation in “non-target field” is, for example, a meaning explanation regarding “soccer team” instead of “candy” for “XX”. Also, the meaning of “ΔΔ” is not “confectionery” but “country”.

Therefore, in this embodiment, in the ranking, the accuracy of the ranking is obtained by using external knowledge information including the meaning explanation in “other than the target field” of the words including the meaning belonging to the target field. To further improve.
That is, in the present embodiment, not only correct answer knowledge information including a description of the meaning of the target character string in the target field but also incorrect knowledge information including a description of the meaning of the target character string outside the target field is acquired. Then, in accordance with the degree of relevance between the plurality of collected information and the correct knowledge information, a rank (first rank) reflecting the correct answer information is determined for each of the collected information. Further, a rank (second rank) in which the incorrect answer knowledge information is reflected is determined for each of the plurality of collected information in accordance with the degree of mutual relation between the plurality of collected information and the incorrect answer knowledge information. Then, based on the difference between the ranking reflecting the correct knowledge information and the ranking reflecting the incorrect knowledge information, the ranking reflecting the correct knowledge information is adjusted, and the final ranking of each of the collected information is determined. To do.

  More specifically, for example, in the order of certain collection information, if the rank reflecting incorrect knowledge information is higher than the rank reflecting correct knowledge information, the rank of the collection information is correct. Adjust it so that it is lower than the ranking that reflects the knowledge information. This is because in such a case, it is estimated that the collected information is likely to be content outside the target field.

  FIG. 3 shows incorrect knowledge information including a meaning explanation of “other than sweets” of “XX” and “ΔΔ” which are target character strings including meanings belonging to “sweets” with respect to the network of FIG. A specific example of a network to which A and incorrect knowledge information B are added is shown. As shown in the specific example, for example, the incorrect answer knowledge information A including the explanation of the meaning of “XX” has a character string such as “soccer” and “team” in common with the collected information B. For this reason, the collection information B has a relatively high degree of association with the incorrect answer knowledge information A. For example, the incorrect answer knowledge information B including the explanation of the meaning of “ΔΔ” has a character string such as “country” or “Japan” in common with the collected information D. For this reason, the collection information D has a relatively high degree of association with the incorrect answer knowledge information B. On the other hand, the collection information A and the collection information C, which are contents related to sweets, have few character strings in common with the incorrect answer knowledge information A and the incorrect answer knowledge information B, and have a relatively low degree of association. In this way, by adding external knowledge information including meaning explanations outside the target field to the network, the degree of relevance of the collected information including contents outside the target field with other collection information becomes relatively high. As a result, as shown in FIG. 2, the rank of the collection information indicating the contents other than the target field is relatively higher than the rank reflecting the external knowledge information including the semantic explanation in the target field. In other words, it is presumed that the collected information indicates contents other than the target field because the ranking reflecting the external knowledge information including the meaning explanation in other than the target field is relatively higher. The For this reason, such collected information is adjusted to lower the rank.

  As described above, in this embodiment, not only external knowledge information including meaning explanation in the target field but also external knowledge information including meaning description in “non-target field” is used for words including meaning belonging to the target field. The order of collected information is adjusted. As a result, the degree to which each piece of information matches the target field can be determined more accurately, and the ranking accuracy can be further improved.

[First embodiment]
<System configuration>
FIG. 4 shows an example of a system configuration in the first embodiment.
This system includes an information collecting server 1, an information providing server 2, and an external knowledge information providing server 3. The information collection server 1, the information providing server 2, and the external knowledge information providing server 3 are connected to be communicable with each other via a network. The network is, for example, a LAN (Local Area Network) or a WAN (Wide Area Network), and may be either a wired connection or a wireless connection. In the present embodiment, the information providing server 2 and the external knowledge information providing server 3 function as a WEB server, and the information collecting server 1 collects various information from the information providing server 2 and the external knowledge information providing server 3 via the Internet. I do.

  The information collection server 1 searches the information providing server 2 for information held in the information providing server 2 with a plurality of target character strings including meanings belonging to the target field, and at least one of the plurality of target character strings. A plurality of collection information 21 including one target character string is collected. In addition, the information collection server 1 receives from the external knowledge information providing server 3 the correct knowledge information 22 including the explanation of the meaning of the target character string in the target field, and the target character string other than the target field. The incorrect answer knowledge information 23 including the meaning explanation is acquired. Then, the information collection server 1 ranks the collected pieces of collected information 21 using the correct answer knowledge information 22 and the incorrect answer knowledge information 23.

The information providing server 2 holds various information in the storage means. In the WEB system, there are generally a plurality of WEB servers such as the information providing server 2, and each holds different information. Each information providing server 2 transmits the data of the collected information 21 to the information collecting server 1 in response to a request from the information collecting server 1.
The external knowledge information providing server 3 holds external knowledge information including explanations of meanings of words in various fields in a storage unit. In response to a request from the information collection server 1, data of external storage information is transmitted to the information collection server 1.

Next, the functional configuration and data configuration of the information collection server 1 will be described in detail.
The information collection server 1 includes an information collection unit 11, a rank determination unit 12, and a rank adjustment unit 13, which realize their functions by installing and executing a program. In addition, the collection means 21, correct answer knowledge information 22, incorrect answer knowledge information 23, correct answer rank table 24, incorrect answer rank table 25, and adjustment rank table 26 are stored in the storage means.

  The information collecting unit 11 searches for information held in the information providing server 2 with a plurality of target character strings including meanings belonging to the target field. Then, the information collection unit 11 collects a plurality of collection information 21 each including at least one target character string from among the plurality of target character strings from the information collection server 1. In addition, the information collection unit 11 provides at least one correct knowledge information 22 including a description of the meaning of the target character string in the target field and an explanation of the meaning of the target character string outside the target field from the external knowledge information providing server 3. At least one incorrect answer knowledge information 23 is acquired.

  The rank determination unit 12 generates a network that indicates the degree of association between pieces of information with the collection information 21 and the correct knowledge information 22 as targets. Then, the rank determining unit 12 determines each rank (first rank) of the collected information 21 according to the degree of relevance between information in the network including the correct knowledge information 22, and stores the correct rank table 24. Store. In addition, the rank determination unit 12 generates a network indicating the degree of association between pieces of information including the incorrect answer knowledge information 23 for the collected information 21. Then, the rank determining unit 12 determines each rank (second rank) of the collection information 21 according to the degree of relevance between the information in the network including the incorrect answer knowledge information 23, and the incorrect answer rank table. 25.

  The rank adjustment unit 13 is based on the rank reflecting the correct answer knowledge information 22 stored in the correct answer rank table 24 and the rank reflecting the incorrect answer knowledge information 23 stored in the incorrect answer rank table 25. The order of the collection information 21 is adjusted. Specifically, the rank adjustment unit 13 adjusts the rank stored in the adjustment rank table 26 and determines each rank of the collection information 21 after adjustment as the final rank.

  The collected information 21 is information collected from the information providing server 2 and is information including at least one target character string among a plurality of target character strings each including a meaning belonging to the target field. The collection information 21 includes, for example, collection information A to collection information D, which are text data as shown in FIG. When the target character string includes meanings other than the target field, the collection information 21 may include information indicating contents other than the target field.

The correct knowledge information 22 is information acquired from the external knowledge information providing server 3 and is information including a description of the meaning of the target character string including the meaning belonging to the target field in the target field. The correct knowledge information 22 includes, for example, correct knowledge information A and correct knowledge information B that are text data as shown in FIG.
The incorrect answer knowledge information 23 is information acquired from the external knowledge information providing server 3 and is information including a description of the meaning of the target character string including the meaning belonging to the target field outside the target field. The incorrect answer knowledge information 23 includes, for example, incorrect answer knowledge information A and incorrect answer knowledge information B, which are text data as shown in FIG.

The correct answer rank table 24 stores the rank of the collected information 21 (that is, the rank reflecting the correct answer knowledge information 22) determined according to the degree of relevance between the collected information in the network including the correct answer knowledge information 22. It is a table. The correct ranking table 24 includes, for example, a ranking and an identifier of the collection information 21 as shown in FIG.
The incorrect answer rank table 25 is the rank of the collection information 21 determined in accordance with the degree of correlation between the collected information in the network including the incorrect answer knowledge information 23 (that is, the rank reflecting the incorrect answer knowledge information 23). Is a table storing. The incorrect answer rank table 25 includes, for example, ranks and identifiers of the collection information 21 as shown in FIG.

The adjustment rank table 26 adjusts the rank reflecting the correct answer knowledge information 22 stored in the correct answer rank table 24 based on the rank reflecting the incorrect answer knowledge information 23 stored in the incorrect answer rank table 25. It is a table for storing the final ranking of the results. The adjustment order table 26 includes, for example, an order and an identifier of the collection information 21 as shown in FIG.
Note that the system configuration shown in FIG. 4 is merely an example. For example, the information collection server 1 can store the collected information 21, the correct answer knowledge information 22 and the incorrect answer knowledge information 23 in advance in a storage unit, instead of receiving them from another server. Further, the information collection server 1 may acquire these pieces of information using a storage medium such as a CD-ROM.

<Description of processing>
The processing executed by the information collection server 1 will be described with reference to the flowcharts shown in FIGS.
FIG. 9 shows the overall processing executed by the information collecting unit 11, the rank determining unit 12 and the rank adjusting unit 13.

  In step S <b> 1, the information collection unit 11 collects, from the information providing server 2, a plurality of collection information 21 including at least one target character string among a plurality of target character strings including meanings belonging to the target field. As this collection method, for example, the information of the information providing server 2 may be searched using the plurality of target character strings, and the search result information may be collected. The collected collection information 21 is stored in the storage means. Note that a plurality of target character strings including meanings belonging to the target field may be acquired, for example, by an input designation by a user, or may be stored in advance in a storage unit or the like.

In step S <b> 2, the information collection unit 11 receives from the external knowledge information providing server 3 correct knowledge information 22 including a description of the meaning of the target character string in the target field, and a description of the meaning of the target character string outside the target field. Incorrect answer knowledge information 23 is acquired.
In addition, as a method for acquiring the correct knowledge information 22 and the incorrect knowledge information 23, for example, the external knowledge information of the external knowledge information providing server 3 is searched with each of a plurality of target character strings, and the search result is obtained. The external knowledge information may be manually distributed to the correct answer knowledge information 22 and the incorrect answer knowledge information 23 by the user. Further, “disambiguation” information obtained when external knowledge information is searched with the above-described target character string may be used. The “disambiguation” information is information that lists a plurality of external knowledge information when the target character string has a plurality of meanings and there are a plurality of pieces of external knowledge information for the target character string. . The specific method of using the “disambiguation” information is as follows. That is, for each of the target character strings, a plurality of pieces of external knowledge information shown in the list obtained by the “disambiguation” are acquired. Then, the degree of relevance between the plurality of external knowledge information is specified. Then, the external knowledge information having the highest degree of association with other external knowledge information is set as correct knowledge information 22 of the target character string, and the other external knowledge information is set as incorrect knowledge information 23 of the target character string. . In this way, automatic distribution to the correct answer knowledge information 22 and the incorrect answer knowledge information 23 becomes possible.

  In step S <b> 3, the rank determining unit 12 generates a network by associating information including the same character string with the collected information 21 and the correct knowledge information 22 as targets. Then, the rank determining unit 12 collects the collected information 21 having a high degree of association with other information based on the degree of association between information in the network, that is, the network reflecting the correct knowledge information 22. Ranking is performed so that 21 is higher. Then, the rank determining unit 12 stores the determined rank in the correct answer rank table 24.

  In step S4, the rank determining unit 12 generates a network by associating information including the same character string with the collected information 21 and the incorrect answer knowledge information 23 as targets. Then, the rank determination unit 12 collects each of the collected information 21 with a high degree of association with other information based on the degree of association between the information in the network, that is, the network reflecting the incorrect answer knowledge information 23. Ranking is performed so that the information 21 is higher. Then, the rank determining unit 12 stores the determined rank in the incorrect answer rank table 25.

  In step S5, the rank adjustment unit 13 obtains the collection information 21 and the incorrect answer knowledge information 23 determined in S4, based on the degree of association determined for the collection information 21 and the correct knowledge information 22 determined in S3. A rank adjustment process is performed for adjustment based on the rank based on the target degree of association. The rank adjustment process will be described in detail next.

FIG. 10 shows details of the order adjustment processing executed by the order adjustment unit 13. This process is executed for each of the collection information 21.
In step S <b> 11, the rank adjustment unit 13 ranks the correct answer knowledge information 22 stored in the correct answer rank table 24 (shown as [A] in the figure) and the incorrect answer rank table 25 stores the incorrect answer rank information 25. The ranks reflecting the correct knowledge information 23 (represented as [B] in the figure) are compared.

  In step S <b> 12, the rank adjustment unit 13 is the same as the rank reflecting the incorrect knowledge information 23 or higher than the rank reflecting the incorrect answer knowledge information 23. Determine. If the condition is met, the process proceeds to step S13 (Yes), and if not, the process proceeds to step S14 (No). In other words, this determination is to determine whether the collection information 21 to be processed includes more (or the same) character strings common to the correct knowledge information 22 than to the incorrect answer knowledge information 23. It is. Note that when the order in which the correct answer knowledge information 22 is reflected is the same as the order in which the incorrect answer knowledge information 23 is reflected, the process may be performed in step S14 described later.

In step S <b> 13, the rank adjustment unit 13 adjusts the rank of the correct answer table 24 as the final rank without adjusting the rank of the collection information 21 to be processed with respect to the rank reflecting the correct knowledge information 22. Store in table 26.
A specific example of the processing in step S13 will be described with reference to FIG. In FIG. 11, the collection information A has the first rank in which the correct knowledge information 22 is reflected. On the other hand, in the collected information A, the ranking reflecting the incorrect answer knowledge information 23 is third. Thus, when the rank reflecting the incorrect answer knowledge information 23 is lower than the rank reflecting the correct answer knowledge information 22, the collected information A is highly likely to be information on the target field. For this reason, the final ranking is used as it is without particularly adjusting the ranking reflecting the correct knowledge information 22.

In step S <b> 14, the rank adjusting unit 13 adjusts the rank of the collection information 21 to be processed so that the rank is lower than the rank reflecting the correct knowledge information 22, and stores it in the adjustment rank table 26. For example, as an example of a calculation method for lowering the rank, the rank of the collection information 21 to be processed is
[Rank reflecting correct knowledge information] + ([Rank reflecting correct knowledge information]-[Rank reflecting incorrect knowledge information])
Adjust so that Then, the adjusted rank is stored in the adjustment rank table 26.

  A specific example of the processing in step S14 will be described with reference to FIG. In FIG. 12, the collection information B has the second rank in which the correct knowledge information 22 is reflected. On the other hand, in the collected information B, the ranking reflecting the incorrect answer knowledge information 23 is first. Thus, when the rank reflecting the incorrect answer knowledge information 23 is higher than the rank reflecting the correct answer knowledge information 22, the collected information B is highly likely to be information other than the target field. For this reason, it adjusts so that the rank may be lowered with respect to the rank reflecting the correct knowledge information 22.

As a result of adjusting the order of all the collected information 21, when the order of the plurality of collected information 21 becomes the same, the same order may be used as it is, or the correct knowledge information 22 is originally reflected. The final ranking may be adjusted so that the highest ranking is the highest ranking. Further, when there is a vacancy in the ranking, the final ranking may be adjusted so that the lower ranking is advanced.
In the above processing, first, the data of the correct answer rank table 24 may be stored as it is in the adjustment rank table 26 and adjusted by changing the rank stored in the adjustment rank table 26. In this case, the rank changes relatively depending on the result of adjusting the rank of the other collected information 21 processed before and after.

<Effects of this embodiment>
According to the present embodiment, the order of the collected information is not simply determined by reflecting the correct knowledge information, but is adjusted in consideration of the order reflecting the incorrect knowledge information. Accuracy is improved.
Specifically, according to the present embodiment, collected information that has more common character strings with incorrect knowledge information than correct knowledge information, that is, collected information that has a high possibility of not conforming to the target field is more accurate. Can be adjusted to lower the ranking. In addition, it is possible to relatively rank the collected information having contents that are likely to be suitable for the target field. If information is selected using this order, information having a meaning suitable for the purpose can be used efficiently.

[Second Embodiment]
<Overview of the second embodiment>
In the first embodiment described above, first, correct knowledge information is added based on the premise that the order of collected information becomes more appropriate by adding external knowledge information in generating a network of collected information. The ranking based on the generated network was determined. Then, the ranking based on the network generated by adding incorrect knowledge information is further determined, and the ranking based on the network generated by adding correct knowledge information is adjusted based on the ranking.

However, if the content of the collected information contains many character strings that are common to all external knowledge information regardless of the field, the accuracy of the ranking may not be improved even if the external knowledge information is reflected. There is.
On the other hand, there are cases where the content of the collected information has few character strings in common with all external knowledge information regardless of the field. As an example of such a case, the sentence expression in the collected information may be a spoken word. In this case, a sentence expression different from the external knowledge information that tends to have many academic phrases is often used, and there are few common character strings.
For this reason, in the second embodiment, in addition to the first embodiment, the ranking is determined in consideration of the ranking determined based only on the collected information without reflecting the external knowledge information. In the following, the description of the same contents as in the first embodiment will be omitted in principle.

<System configuration>
FIG. 13 shows an example of a system configuration in the second embodiment. In the second embodiment, in addition to the system configuration of the first embodiment, the information collection server 1 includes an original rank table 27.
The original rank table 27 is a table that stores the rank (third rank) of the collection information 21 determined according to the degree of association between the collection information in the network of only the collection information 21 (network not including external knowledge information). is there. The original rank table 27 includes, for example, ranks and identifiers of the collection information 21 as shown in FIG.

<Description of processing>
The processing executed by the information collection server 1 in the second embodiment will be described with reference to the flowcharts shown in FIGS.
FIG. 15 shows the overall processing executed by the information collecting unit 11, the rank determining unit 12 and the rank adjusting unit 13.
Steps S21 and S22 are the same as steps S1 and S2 in the first embodiment shown in FIG.

In step S <b> 23, the rank determination unit 12 generates a network by associating information including the common character string with the collected information 21 as a target. Then, the rank determination unit 12 ranks information having a high degree of association with other information based on the degree of association between information in the network. Then, the rank determining unit 12 stores the determined rank in the original rank table 27.
Steps S24 to S26 are the same as steps S3 to S5 in the first embodiment shown in FIG.

FIG. 16 shows details of the order adjustment processing executed by the order adjustment unit 13 in the second embodiment. This process is executed for each of the collection information 21.
In step S31, the rank adjustment unit 13 stores the correct answer knowledge information stored in the correct answer rank table 24, which is stored in the original rank table 27 and does not reflect the external knowledge information (indicated by [X] in the figure). The rank reflecting 22 (represented as [Y] in the figure) and the rank reflecting the incorrect answer knowledge information 23 stored in the incorrect answer rank table 25 (represented as [Z] in the figure) are compared.

  In step S32, the rank adjusting unit 13 determines that both the rank reflecting the correct knowledge information 22 and the rank reflecting the incorrect knowledge information 23 are higher than the rank not reflecting the external knowledge information, or the correct answer. It is determined whether both the rank reflecting the knowledge information 22 and the rank reflecting the incorrect answer knowledge information 23 are lower than the rank not reflecting the external knowledge information. If any of these conditions is met, the process proceeds to step S33 (Yes), and if not, the process proceeds to step S34 (No).

In step S <b> 33, the rank adjustment unit 13 stores the rank that does not reflect the external knowledge information in the adjustment rank table 26.
A specific example of the processing in step S33 will be described with reference to FIGS. In FIG. 17, the collected information B is ranked 4th in the order not reflecting external knowledge information. On the other hand, the rank reflecting the correct answer knowledge information 22 is second, and the rank reflecting the incorrect answer knowledge information 23 is first. As described above, when both the rank reflecting the correct knowledge information 22 and the rank reflecting the incorrect answer knowledge information 23 are higher than the rank not reflecting the external knowledge information, the collected information B is the external knowledge information. Although it is presumed that the information includes many character strings that are common to the whole, it is not clear in processing whether the information is in the target field. For this reason, it is avoided to adjust the rank that does not reflect the external knowledge information, and the rank that does not reflect the external knowledge information is used as the determination rank.

  Similarly, in FIG. 18, the collection information C is second in the order not reflecting external knowledge information. On the other hand, the rank reflecting the correct answer knowledge information 22 is third, and the rank reflecting the incorrect answer knowledge information 23 is fourth. As described above, when both the rank reflecting the correct answer knowledge information 22 and the rank reflecting the incorrect answer knowledge information 23 are lower than the rank not reflecting the correct answer knowledge information 22, the collected information C is stored in the external knowledge. Although it is presumed that the information has few character strings in common with the information in general, it is not clear in processing whether the information is in the target field. For this reason, the order that does not reflect the external knowledge information is used as the determination order.

In step S34, the rank adjustment unit 13 determines whether the rank reflecting the incorrect knowledge information 23 is higher than the rank not reflecting the external knowledge information. If the condition is met, the process proceeds to step S35 (Yes), and if not, the process proceeds to step S36 (No).
In step S <b> 35, the rank adjustment unit 13 performs adjustment so that the rank of the collection information 21 to be processed is lower than the rank that does not reflect the external knowledge information. For example, as an example of a calculation method for lowering the rank, the rank of the collection information 21 to be processed is
[Order for not reflecting external knowledge information]-([Order for reflecting incorrect knowledge information]-[Order for not reflecting external knowledge information])
Adjust so that Then, the adjusted rank is stored in the adjustment rank table 26.

  A specific example of the process of step S35 will be described with reference to FIG. In FIG. 19, the collected information B is ranked second in the order not reflecting external knowledge information. On the other hand, in the collected information B, the ranking reflecting the incorrect answer knowledge information 23 is first. As described above, when the rank in which the incorrect knowledge information 23 is reflected is higher than the rank in which the external knowledge information is not reflected, the collected information B is highly likely to be information other than the target field. For this reason, it adjusts so that a rank may be lowered with respect to a rank that does not reflect external knowledge information.

  In step S <b> 36, the rank adjustment unit 13 determines whether the rank reflecting the incorrect knowledge information 23 is lower than the rank not reflecting the external knowledge information. If the condition is met, the process proceeds to step S37 (Yes), and if not, the process proceeds to step S38 (No).

In step S <b> 37, the rank adjustment unit 13 performs adjustment so that the rank of the collection information 21 to be processed is higher than the rank that does not reflect the external knowledge information. For example, as an example of a calculation method for raising the rank, the rank of the collection information 21 to be processed is
[Order not reflecting external knowledge information] + ([Order not reflecting external knowledge information]-[Order reflecting incorrect knowledge information])
Adjust so that Then, the adjusted rank is stored in the adjustment rank table 26.

  A specific example of the processing in step S37 will be described with reference to FIG. In FIG. 20, the collection information A is ranked second in the order not reflecting external knowledge information. On the other hand, in the collected information A, the ranking reflecting the incorrect answer knowledge information 23 is third. As described above, when the order in which the incorrect knowledge information 23 is reflected is lower than the order in which the external knowledge information is not reflected, there is a high possibility that the collected information A is information in the target field. For this reason, it adjusts so that a rank may be raised with respect to the rank which does not reflect external knowledge information.

  In step S <b> 38, the rank adjustment unit 13 stores the rank reflecting the correct knowledge information 22 in the adjustment rank table 26. The case where step S38 is performed is a case where the order in which the incorrect knowledge information 23 is reflected does not change from the order in which the external knowledge information is not reflected.

In step S35 and step S37, the order in which the external knowledge information is not reflected is adjusted based on the order in which the incorrect knowledge information 23 is reflected, but the correct knowledge information 22 is reflected as in step S38. You may make it store the made order | rank in the adjustment order | rank table 26. FIG.
In the above process, for example, the data of the original rank table 27 or the data of the correct answer rank table 24 is first stored in the adjustment rank table 26 as it is, and the rank stored in the adjustment rank table 26 is changed and adjusted. It may be.

<Effects of this embodiment>
According to the present embodiment, for each collection information, it can be estimated whether or not it is appropriate to adjust the rank based on the result of reflecting the external knowledge information. Then, when it is not appropriate to adjust the order, the order based on the network generated without adding external knowledge information can be applied. For this reason, it is possible to avoid making unnecessary adjustments as compared with the case where the order is adjusted by reflecting correct knowledge information uniformly.

[Hardware configuration, etc.]
An example of the hardware configuration of the computer functioning as the information collection server 1 described above is shown in FIG. The computer includes a processor 101, a memory 102, a storage 103, a portable storage medium drive device 104, an input / output device 105, and a communication interface 106.
The processor 101 includes a control unit, an arithmetic unit, an instruction decoder, and the like. The execution unit follows the instructions of the program decoded by the instruction decoder, and performs arithmetic / logic using the arithmetic unit according to a control signal output from the control unit. Perform the operation. The processor 101 has a TLB that functions as a control register in which various information used for control is stored, a cache that can temporarily store the contents of the already accessed memory 2 and the like, and a page table cache of virtual memory. Prepare. The processor 101 may have a configuration in which a plurality of CPU (Central Processing Unit) cores are provided.

  The memory 102 is a storage device such as a RAM (Random Access Memory), for example, and is a main memory in which a program to be executed by the processor 101 is loaded and data used for processing of the processor 101 is stored. The storage 103 is a storage device such as an HDD (Hard Disk Drive) or a flash memory, and stores programs and various data. The portable storage medium driving device 104 is a device that reads data and programs stored in the portable storage medium 107. The portable storage medium 107 is, for example, a magnetic disk, an optical disk, a magneto-optical disk, or a flash memory. The processor 101 executes a program stored in the storage 103 or the portable storage medium 107 while cooperating with the memory 102 or the storage 103. Note that the program executed by the processor 101 and data to be accessed may be stored in another device that can communicate with the computer. Note that the storage means of the information collection server 1 described in the present embodiment indicates at least one of the memory 102, the storage 103, the portable storage medium 107, or another device that can communicate with the computer.

The input / output device 105 is, for example, a keyboard, a touch panel, a display, or the like, and receives an operation command by a user operation or the like, and outputs a processing result by a computer.
The communication interface 106 can include, for example, a radio frequency receiver and transmitter, and an optical receiver and transmitter in addition to a LAN (Local Area Network) card, for example. The aforementioned receivers and transmitters can be implemented to operate with one or more communication networks, such as a Wi-Fi network, a Bluetooth network, and long term evolution.
Each component of these computers is connected by a bus 108.

[Others]
Note that the functional configuration and physical configuration of the computer described in this specification are not limited to the above-described aspects. For example, the functions and physical resources are integrated and implemented, or conversely, are further distributed. It is also possible to implement.
In addition, in this specification, the description places “to be more than” and “to be less than” in comparison with the threshold value and the like are not limited to the description except for special cases, It can be appropriately replaced with “less than” or “less than (less than)”. The reverse is also true.

Regarding the above embodiment, the following additional notes are disclosed.
(Appendix 1)
Out of a plurality of target character strings including meanings belonging to a predetermined target field, a plurality of collected information each including at least one target character string, and external knowledge information including a description of the meaning of the target character string in the target field Correct knowledge information that is, and incorrect knowledge information that is external knowledge information including a description of the meaning of the target character string outside the target field,
In accordance with the degree of relevance between the plurality of collection information and the correct answer knowledge information, the first ranking of each of the plurality of collection information is determined,
In accordance with the degree of relevance between the plurality of collected information and the incorrect knowledge information, a second order of each of the plurality of collected information is determined,
An information collection program that adjusts the first rank based on the difference between the first rank and the second rank, and causes a computer to execute a process of determining the final rank of each of the plurality of pieces of collected information.

(Appendix 2)
In the process of determining the final rank, when the first rank of the collected information is lower than the second rank, the rank of the collected information is further lowered than the first rank. The information collection program according to appendix 1, wherein adjustment is performed so that

(Appendix 3)
Further including a process of determining a third rank of each of the plurality of pieces of collection information in accordance with a degree of mutual relevance in the plurality of pieces of collection information;
The process of determining the final rank is performed when both the first rank and the second rank are higher than the third rank, or the first rank and the second rank The information collection program according to attachment 1 or 2, wherein when both are lower than the third rank, the third rank is the final rank.

(Appendix 4)
When the plurality of pieces of external knowledge information exist for each of the target character strings, the acquiring process is performed with other external knowledge information included in the plurality of external knowledge information. The external knowledge information having the highest degree of relevance is acquired as the correct answer knowledge information, and external knowledge information other than the correct answer knowledge information is acquired as the incorrect answer knowledge information. Information gathering program.

(Appendix 5)
Out of a plurality of target character strings including meanings belonging to a predetermined target field, a plurality of collected information each including at least one target character string, and external knowledge information including a description of the meaning of the target character string in the target field Correct knowledge information that is, and incorrect knowledge information that is external knowledge information including a description of the meaning of the target character string outside the target field,
In accordance with the degree of relevance between the plurality of collection information and the correct answer knowledge information, the first ranking of each of the plurality of collection information is determined,
In accordance with the degree of relevance between the plurality of collected information and the incorrect knowledge information, a second order of each of the plurality of collected information is determined,
An information collection method in which a computer executes a process of adjusting the first rank based on a difference between the first rank and the second rank and determining a final rank of each of the plurality of pieces of collected information.

(Appendix 6)
Out of a plurality of target character strings including meanings belonging to a predetermined target field, a plurality of collected information each including at least one target character string, and external knowledge information including a description of the meaning of the target character string in the target field Correct acquisition knowledge information, and an information acquisition unit for acquiring incorrect knowledge information that is external knowledge information including explanation of the meaning of the target character string outside the target field;
According to the degree of relevance between the plurality of collection information and the correct knowledge information, a first rank of each of the plurality of collection information is determined, and the plurality of collection information and the incorrect answer knowledge information A rank determining unit that determines a second rank of each of the plurality of pieces of collection information in accordance with a degree of mutual relevance;
An information collection apparatus comprising: a rank adjustment unit that adjusts the first rank based on a difference between the first rank and the second rank and determines a final rank of each of the plurality of collection information.

DESCRIPTION OF SYMBOLS 1 ... Information collection server, 2 ... Information provision server, 3 ... External knowledge information provision server, 11 ... Information collection part, 12 ... Order determination part, 13 ... Order adjustment part, 21 ... Collection information, 22 ... Correct knowledge information, 23 ... incorrect answer knowledge information, 24 ... correct answer rank table, 25 ... incorrect answer rank table, 26 ... adjustment rank table, 27 ... original rank table

Claims (6)

Out of a plurality of target character strings including meanings belonging to a predetermined target field, a plurality of collected information each including at least one target character string, and external knowledge information including a description of the meaning of the target character string in the target field Correct knowledge information that is, and incorrect knowledge information that is external knowledge information including a description of the meaning of the target character string outside the target field,
In accordance with the degree of relevance between the plurality of collection information and the correct answer knowledge information, the first ranking of each of the plurality of collection information is determined,
In accordance with the degree of relevance between the plurality of collected information and the incorrect knowledge information, a second order of each of the plurality of collected information is determined,
An information collection program that adjusts the first rank based on the difference between the first rank and the second rank, and causes a computer to execute a process of determining the final rank of each of the plurality of pieces of collected information.   In the process of determining the final rank, when the first rank of the collected information is lower than the second rank, the rank of the collected information is further lowered than the first rank. The information collection program according to claim 1, wherein adjustment is performed so as to perform. Further including a process of determining a third rank of each of the plurality of pieces of collection information in accordance with a degree of mutual relevance in the plurality of pieces of collection information;
The process of determining the final rank is performed when both the first rank and the second rank are higher than the third rank, or the first rank and the second rank The information collection program according to claim 1 or 2, wherein when both are lower than the third rank, the third rank is the final rank.   When the plurality of pieces of external knowledge information exist for each of the target character strings, the acquiring process is performed with other external knowledge information included in the plurality of external knowledge information. The external knowledge information having the highest degree of relevance is acquired as the correct answer knowledge information, and external knowledge information other than the correct answer knowledge information is acquired as the incorrect answer knowledge information. Information gathering program. Out of a plurality of target character strings including meanings belonging to a predetermined target field, a plurality of collected information each including at least one target character string, and external knowledge information including a description of the meaning of the target character string in the target field Correct knowledge information that is, and incorrect knowledge information that is external knowledge information including a description of the meaning of the target character string outside the target field,
In accordance with the degree of relevance between the plurality of collection information and the correct answer knowledge information, the first ranking of each of the plurality of collection information is determined,
In accordance with the degree of relevance between the plurality of collected information and the incorrect knowledge information, a second order of each of the plurality of collected information is determined,
An information collection method in which a computer executes a process of adjusting the first rank based on a difference between the first rank and the second rank and determining a final rank of each of the plurality of pieces of collected information. Out of a plurality of target character strings including meanings belonging to a predetermined target field, a plurality of collected information each including at least one target character string, and external knowledge information including a description of the meaning of the target character string in the target field Correct acquisition knowledge information, and an information acquisition unit for acquiring incorrect knowledge information that is external knowledge information including explanation of the meaning of the target character string outside the target field;
According to the degree of relevance between the plurality of collection information and the correct knowledge information, a first rank of each of the plurality of collection information is determined, and the plurality of collection information and the incorrect answer knowledge information A rank determining unit that determines a second rank of each of the plurality of pieces of collection information in accordance with a degree of mutual relevance;
An information collection apparatus comprising: a rank adjustment unit that adjusts the first rank based on a difference between the first rank and the second rank and determines a final rank of each of the plurality of collection information.