JP5032645B2 - News information analyzer - Google Patents

Description

  Embodiments described herein relate generally to a news information analysis apparatus.

  2. Description of the Related Art Conventionally, as a technique for determining information provision priority (importance) of Internet distribution news, there is a technique that utilizes the newness of news and the number of related news. In this technology, for example, the degree of association is obtained from the number of word matches between the delivered news and other information, the freshness of the news is obtained from the elapsed time from the news delivery time, and the priority of information provision is determined.

  In addition, in order to support the grasp of fluctuations in stock prices and news that has affected stock prices, for example, a technique for displaying a stock price chart and related news on the same screen is known.

JP 2001-209655 A Japanese Patent No. 3300366 JP 2003-108785 A

  However, the techniques as described above are usually not particularly problematic, but have various disadvantages. According to the study of the present inventor, this inconvenience is caused by the low accuracy of identity determination for a plurality of distributed news information, or not performing identity determination. In the present specification, the term “identity” may be appropriately replaced with the terms “relevance”, “match”, or “similarity”.

  For example, in the technology for determining the information provision priority, when the degree of relevance is calculated from the number of word matches, the difference in distribution time between the distributed news and other information is not taken into consideration. There is an inconvenience that the distinction of “Product Y additional recall at Company A” is insufficient. Moreover, since the sameness between the numerical information of the distributed news and the numerical information of other information is not considered, there is a disadvantage that it is difficult to cope with information correction.

  Further, in the technology for displaying the stock price chart and the related news on the same screen, the news identity determination is not executed. For this reason, since a large number of news items that are identical to the actual contents of the report (reported facts) are displayed, there is a disadvantage that it is difficult to grasp the whole picture.

  The problem to be solved by the present invention is to provide a news information analysis apparatus capable of performing identity determination on a plurality of distributed news information and improving the accuracy of identity determination.

  The news information analysis device of the embodiment can communicate with both the news distribution site device and the user terminal device individually.

  When news data including a title described in Japanese is distributed from a news source device of a news source, the news distribution site device of the embodiment distributes the date and time of the distribution or the date and time when it was posted on the news distribution site. Transmitting news information including a date and time, a distribution site name indicating the news distribution site, a news source name indicating the news source, and a title in which the title of the news data is the same as the title of the news data or the title is replaced. The news distribution site device of the news distribution site.

  The user terminal device according to the embodiment transmits the analysis result of the news information.

  The news information analysis apparatus of the embodiment includes a memory, an evaluation word dictionary storage unit, an identity determination criterion storage unit, an importance calculation procedure storage unit, a news reception unit, a morpheme analysis unit, an evaluation result writing unit, an identity determination unit, Importance calculation means and analysis result transmission means are provided.

  The evaluation word dictionary storage means of the embodiment stores an evaluation word dictionary in which the corresponding expression, heading expression, major classification, and minor classification are associated with each other.

  The corresponding expression indicates a regular expression of a morphological analysis result for a term including one or more words that may be included in the title of the news information.

  The heading expression indicates a heading corresponding to the corresponding expression (a synonym of one or a plurality of corresponding expressions, a label that summarizes notation fluctuations).

  The minor classification indicates that the meaning of the headline expression is a company, or forward or backward.

  The major classification indicates that the meaning of the minor classification is object or evaluation.

  The identity determination criterion storage means of the embodiment stores an identity determination criterion indicating a criterion for determining whether or not any two pieces of news information in the transmitted news information are the same news.

  The identity determination criterion of the embodiment is that the news source names included in the two news information matches each other, and the distribution time difference indicating the difference in distribution date / time included in the two news information is more than the reference value. The similarity calculated from the independent words and the numerical information in the word analysis results of the two titles in the two news information is higher than the specified value, and the similarity is a significant number of the numerical information Calculated after rounding to the nearest whole number.

  The importance calculation procedure storage unit of the embodiment stores the importance calculation procedure.

  The importance calculation procedure includes calculating the importance of the news information based on the number of news information determined to be the same news and the similarity calculated from the word analysis result when the negative is determined. The procedure for calculating is shown.

  The news receiving means of the embodiment, when receiving news information from the news distribution site device, adds a news ID to the news information and writes the obtained ID-added news information in the memory.

  The word analysis unit of the embodiment analyzes the title included in the ID-added news information in the memory, adds the obtained word analysis result to the news ID and title of the ID-added news information, and obtains the analysis Result information is written into the memory.

  The evaluation result writing means of the embodiment extracts a term corresponding to the corresponding expression from the word analysis result in the analysis result information in the memory, and the major classification and the minor classification related to the corresponding expression are extracted from the word The evaluation result is extracted from the evaluation word dictionary, and the evaluation result including the extracted term, the large classification, and the small classification is associated with the news ID that matches the news ID of the analysis result information and written into the memory.

  The identity determination unit according to the embodiment includes the latest ID-added news information in the memory and the ID-added news information written in the past based on whether or not the identity determination criterion in the identity determination criterion storage unit is satisfied. Are the same news.

  The importance calculation unit according to the embodiment calculates the importance of the latest ID-added news information based on the importance calculation procedure in the importance calculation procedure storage unit and the determination result by the identity determination unit. .

  The analysis result transmission means of the embodiment reads the evaluation result associated with the news ID that matches the news ID in the latest ID-added news information from the memory, the read evaluation result, and the calculated importance Is sent to the user terminal device.

It is a block diagram which shows the structural example of the news information analysis system which concerns on 1st Embodiment. It is a schematic diagram which shows an example of the news information in the embodiment. It is a schematic diagram which shows an example of the evaluation word dictionary information in the embodiment. It is a schematic diagram which shows an example of news information with ID in the embodiment. It is a schematic diagram which shows the example of the analysis result information in the same embodiment. It is a schematic diagram which shows the example of the analysis result information in the same embodiment. It is a schematic diagram which shows the example of the analysis result information in the same embodiment. It is a schematic diagram which shows the example of the analysis result information in the same embodiment. It is a schematic diagram which shows the example of the evaluation result and news ID in the embodiment. It is a schematic diagram which shows an example of the analysis result in the same embodiment. It is a flowchart for demonstrating operation | movement of the news information analysis system in the embodiment. It is a schematic diagram which shows an example of news information with ID in the embodiment. It is a schematic diagram for demonstrating the independent word extracted from the morphological analysis result in the analysis result information in the embodiment. It is a schematic diagram for demonstrating the numerical information and independent word in the same embodiment. It is a schematic diagram which shows an example of the evaluation word dictionary memorize | stored in the memory | storage part of the news information analyzer which concerns on 2nd Embodiment. It is a schematic diagram which shows an example of the extraction result by the information evaluation part in the embodiment. It is a schematic diagram for demonstrating the independent word extracted from the morphological analysis result in the analysis result information in the embodiment. It is a schematic diagram which shows an example of the analysis result in the same embodiment. It is a schematic diagram for demonstrating the independent word extracted from the morphological analysis result in the analysis result information in 3rd Embodiment. It is a schematic diagram which shows an example of the statistical information in 5th Embodiment. It is a schematic diagram which shows an example which linked | related and stored the delivery site name and news source name, and the reliability in 6th Embodiment.

  Each embodiment will be described below with reference to the drawings. Each of the following devices can be implemented for each device with either a hardware configuration or a combination configuration of hardware resources and software. As the software of the combined configuration, a program that is installed in advance on a computer of a corresponding device from a network or a storage medium and that realizes the function of the corresponding device is used. In addition, as registered trademarks in the following description, there are Nikkei 225 (registered trademark) and TSE (registered trademark) codes.

<First Embodiment>
FIG. 1 is a block diagram illustrating a configuration example of the news information analysis system according to the first embodiment, and FIGS. 2 to 10 are schematic diagrams illustrating an example of each piece of information. The news information analysis system is configured such that a plurality of news information distributed from a news distribution site device (not shown) is analyzed by the news information analysis device 10 and the analysis result of the news information is transmitted to the user terminal device 20. .

  The news distribution site device has a function of posting news data on a news distribution site when news data including a title described in Japanese is distributed from a news source device (not shown) of the news source, and the news data And the function of transmitting news information based on the news information analyzing apparatus 10 to the news information analyzing apparatus 10. As shown in FIG. 2, the news information includes a time stamp as a date and time when news data is distributed or a date and time when the data is posted on a news distribution site, a distribution site name indicating the news distribution site, It includes a news source name indicating a news source, and a title (denoted as “news” in the figure) that is the same title as the title of the news data or a word in which the title is replaced.

  On the other hand, the news information analysis apparatus 10 includes a storage unit 11, a news reception unit 12, a word analysis unit 13, an information evaluation unit 14, an identity determination unit 15, an importance level calculation unit 16, a control unit 17, and an input / output interface 18. 19 is connected.

  The storage unit (memory, evaluation word dictionary storage means, identity determination reference storage means, and importance calculation procedure storage means) 11 is configured by hardware such as a hard disk drive or a nonvolatile memory device. The storage unit 11 includes a program executed by the control unit 17, a word analysis dictionary used by the word analysis unit 13, an evaluation word dictionary used by the information evaluation unit 14, an identity determination criterion used by the identity determination unit 15, and an importance level. The importance calculation procedure and news information used in the calculation unit 16 are stored.

  Here, as shown in FIG. 3, the evaluation word dictionary indicates a regular expression related to a result of word analysis such as morphological analysis for a term including one or more words that may be included in the title of news information. “Heading expression”, “Heading expression” indicating the title corresponding to this “Heading expression”, and the meaning of this “Heading expression” is, for example, a company, or the meaning of evaluation is positive or positive (Negative) “minor classification”, and the meaning of this “minor classification” is, for example, information that associates the target or “major classification” that indicates evaluation. Yes.

  Supplementally, “major classification” indicates, for example, whether the information extracted as an example is “target” or “evaluation”. “Small classification” indicates the type of information extracted as an example. For example, for the major category “object”, there are minor categories “company” and “local government”, and for the major category “evaluation”, there are minor categories “negative” and “positive”. Of course, the “major classification” and the “small classification” are not limited to these examples, and can be set according to the needs of the user. That is, “minor classification” indicates the meaning of the heading expression, and “major classification” indicates the meaning of “small classification”, that is, the distinction of the heading expression.

  “Heading expression” is an item name of information extracted by the corresponding line of the evaluation word dictionary. The “corresponding expression” refers to a regular expression pattern of the word analysis result registered in the evaluation word dictionary as a corresponding expression to be extracted from the news title after word analysis. This regular expression is a method for expressing a character string using a specific character (metacharacter). For example, “^” is the beginning of the line, “$” is the end of the line, “.” Is any single character, “(|)” is one of the expressions delimited by “|”, “[]” is inside Select one of the arranged characters, “¥” means that the next character is not interpreted as a metacharacter.

  To summarize, the hierarchy is formed in the order of “major classification”> “minor classification”> “heading expression”> “corresponding expression”, and each is defined by a relationship of 1: n (n is an integer of 1 or more). . That is, one or more “minor categories” belong to one of the “major categories”, one or more “headline expressions” belong to one of the “minor categories”, and one of the “headline expressions” One or more “corresponding expressions” belong to it.

  Although only the corresponding expression is described in the example of FIG. 3, in addition to the corresponding expression, an “exclusion expression” that is defined as an expression that should be excluded even if the corresponding corresponding expression is matched may be defined. In addition, when there is little risk of erroneous word division without performing word analysis such as morphological analysis, a regular expression pattern of an expression that is not subjected to word analysis may be set as “corresponding expression”.

  The identity determination criterion indicates a criterion for determining whether or not any two pieces of news information in the transmitted news information are the same news, and the news source names included in the two pieces of news information are mutually different. Match, the difference in distribution time indicating the difference in distribution date and time included in the two news information items is smaller than the reference value (maximum distribution time difference), and the morphemes of the two titles in the two news information items This includes that the similarity calculated from the analysis result is higher than a specified value, and that the similarity is calculated after rounding the digits of significant digits of the numerical information.

  The similarity is included in both the morphological analysis results of the two titles with respect to the number of independent words and numerical information in the entire set of independent words included in the morphological analysis results of the two titles, for example. It is the ratio of the number of independent words and numerical information. Further, the specified value of the similarity (0 or more and 1 or less) is preferably a high value of about 0.9. Also, instead of “the similarity calculated from the morphological analysis results of the two titles is higher than the specified value”, “the free words extracted from the morphological analysis results of the two titles match each other” Also good. The identity determination criterion may be read as an identity determination rule.

  The importance calculation procedure includes the number of news information determined to be the same news, and the similarity calculated from the morphological analysis result when the identity determination determines NO (when it is determined that the news is not the same). The procedure for calculating the importance of the news information based on the above is shown. Further, the importance calculation procedure may be read as the importance determination rule.

  When receiving news information from the news distribution site device, the news receiving unit 12 has a function of adding a news ID to the news information and writing the obtained ID-added news information in the storage unit 11. The news information may be received in a form in which a contract is made with a distributor and the distribution is received, or a news that is newly distributed by an Internet search program generally called a search robot or a crawler is found and obtained.

  Here, as shown in FIG. 4, the ID-added news information includes four items of “time stamp”, “delivery site name”, “news source name”, “news” in the news information, and “news” added by the news receiving unit 12. ID ”item. “Time stamp” is distribution date information of news information. The “distribution site name” is the name of the news distribution site corresponding to the news distribution site apparatus that has transmitted the news information to the apparatus 10. The “news source name” is the name of the company that distributed the news data to the distribution site “distribution site name”. “Distribution site name” and “news source name” may match. “News” is a Japanese string of the title portion of the distributed news information.

  The word analysis unit 13 uses the word analysis dictionary in the storage unit 11 to perform word analysis processing of the news information with ID. For example, the word analysis process uses a morphological analysis technique (a known technique). In other words, the word analysis unit 13 performs morphological analysis on the title included in the ID-added news information in the storage unit 11, and adds the obtained morpheme analysis result to the news ID and title of the ID-added news information. A morpheme analysis function for writing the analysis result information to the storage unit 11.

  In the following description, the morphological analysis technique will be described as an example. However, the processing in the word analysis unit 13 may be performed by using an analysis process that does not use a word analysis dictionary or is not a morphological analysis such as an N-gram and is decomposed into words. Good. That is, according to the present invention, news information is divided into words by a technique not limited to morphological analysis, and similarity is determined by comparing words.

  However, for example, in the case of an N-gram with a character as a unit, it is not possible to determine whether it is part-of-speech or independent words. Therefore, when word analysis processing is performed by other means of morphological analysis, “words” can be processed instead of “independent words”.

  The analysis result information includes the news (title) in the ID-added news information broken down into words as shown in FIGS. 5 to 8 in which an example of the analysis result by the morphological analysis free software “tea bowl” is shown. As shown in the figure, the information in the “news” column is broken down into words by the symbol “/”, and part-of-speech information is given in “<>”.

  The information evaluation unit 14 extracts a term corresponding to the “corresponding expression” from the morphological analysis result in the analysis result information in the storage unit 11, and “major classification” and “small classification” related to the “corresponding expression”. Is extracted from the evaluation word dictionary, and the evaluation result including the extracted term, the large classification, and the small classification is associated with the news ID that matches the news ID of the analysis result information and written in the storage unit 11.

  For example, the information evaluation unit 14 uses the expression “Company A <noun-proper noun-” to which the classification of the major classification “target”, minor classification “company” is given from the morphological analysis result in the analysis result information including the news ID “38”. The expression “recall <noun-sa change connection>” with the classification “organization>”, major classification “evaluation”, minor classification “negative”, etc. is extracted, and the extracted company name “Company A” as shown in FIG. The evaluation result including the evaluation “negative”, the evaluation specific word “recall” and the like is written in the storage unit 11 in association with the news ID “38”. The item name “evaluation concrete term” is a name of a subordinate concept of the item name “term”.

  Based on whether the identity determination criteria in the storage unit 11 are satisfied, the identity determination unit 15 determines that the latest ID-added news information in the storage unit 11 and the ID-added news information written in the past are the same news. Has a function to determine whether or not there is.

  For example, the identity determination unit 15 extracts the independent words and numerical information extracted from the morphological analysis result of the title in the latest ID-added news information and the morphological analysis result of the title in the ID-added news information written in the past. The degree of similarity indicating the ratio that the independent word and the numerical information match is calculated. However, before calculating the similarity, the digits of the significant digits in the numerical information are rounded off. When the similarity is higher than a specified value (eg, 0.9), the identity determination unit 15 matches the news source names in the ID-added news information, and further, the difference in the distribution time of each news is a reference value ( For example, if it is within 5 minutes, it is determined that the news is the same. Note that confirmation that the degree of similarity is high, news source name match confirmation, and delivery time difference confirmation can be performed in any order. Further, instead of confirming that the degree of similarity is high, it may be confirmed that the independent words completely match.

  The importance calculation unit 16 includes an importance calculation function for calculating the importance of the latest ID-added news information based on the importance calculation procedure in the storage unit 11 and the determination result by the identity determination unit 15, and the latest The evaluation result associated with the news ID that matches the news ID in the ID-added news information is read from the storage unit 11, and the analysis result including the read evaluation result and the calculated importance is input to the user by the input / output interface 18. It has an analysis result transmission function to be transmitted to the terminal device 20. The importance calculation procedure is, for example, “the number of the same news in the past is regarded as the importance”, “similar news with a past similarity of 0.8 or more is added to the importance of the latest received news”, etc. The procedure for calculating the importance is described. As shown in FIG. 10, for example, in the analysis result, importance is added to data obtained by deleting a news ID from the above-described evaluation result. Note that the analysis result may include a news ID.

  The control unit 17 has a function of controlling the units 12 to 16 and 18 by executing a program in the storage unit 11.

  The input / output interface 18 is used as a communication interface for the user terminal device 20. However, the input / output interface 18 can be connected to an external storage device (not shown) via a cable, and a dictionary and data may be input / output to / from the storage unit 11 with the external storage device.

  The user terminal device 20 is a terminal device handled by a user who uses the news information analysis device 10. In the user terminal device 20, a terminal control unit 21, a terminal storage unit 22, a display unit 23, an input unit 24, and a communication interface 25 that control each unit 23 to 25 are connected via a bus.

  The terminal storage unit 22 is a storage device configured by hardware such as a hard disk drive or a non-volatile memory device, and stores a control program. The terminal storage unit 22 also functions as a work memory for various processes performed by the terminal control unit 21.

  The display unit 23 is a liquid crystal display, for example, and has a function of displaying the analysis result received from the news information analysis apparatus 10. The display unit 23 displays, for example, the distribution time of the newly received news, the distribution site, the news source, the company name / industry that has been discussed in the news, the type of news (positive, negative, unstable, etc.), and the importance level. Present to users in tabular form.

  The input unit 24 is, for example, a keyboard or a mouse, and receives input related to news information analysis and system settings in accordance with a user operation.

  The communication interface 25 is connected to the news information analysis apparatus 10 via a cable, and transmits / receives to / from the news information analysis apparatus 10 a specified condition received by the input unit 24 and an analysis result of the news information analysis apparatus 10. . The communication interface 25 can be connected to an external storage device via a cable. A word analysis dictionary, an evaluation word dictionary, an identity determination criterion, and an importance calculation procedure stored in the storage unit 11 with the external storage device. Analysis results can also be input and output.

  Next, the operation of the news information analysis system configured as described above will be described using the flowchart of FIG.

  First, the news receiving unit 12 receives a distribution from the news distribution site device or accesses the news distribution site, and thereby, for example, includes a new time stamp “2010/2/4 9:53” shown in FIG. When the news information is received (step S1), for example, a news ID “191” is added to the news information, and the obtained news information with ID is written in the storage unit 11, for example, as shown in FIG.

  The word analysis unit 13 performs word analysis on the “news” portion included in the ID-added news information in the storage unit 11 (step S2), and adds the obtained morpheme analysis result to the news ID and title of the ID-added news information. Then, the obtained analysis result information is written in the storage unit 11. For example, the news (title) of the news ID “191” is analyzed as shown in the “morpheme analysis result” data associated with the news ID “191” shown in FIG.

  Subsequently, the information evaluation unit 14 extracts a term corresponding to the “corresponding expression” from the morphological analysis result in the analysis result information in the storage unit 11, and “major classification” and “ "Small category" is extracted from the evaluation word dictionary, and the evaluation result including the extracted term, major category and minor category is written in the storage unit 11 in association with the news ID matching the news ID of the analysis result information (step S3). ).

  In step S3, for example, the information evaluation unit 14 performs a pattern matching process for the corresponding expression in each row of the evaluation word dictionary shown in FIG. 3 on the morphological analysis result in the analysis result information in the storage unit 11. If the corresponding expression matches the morphological analysis result in the analysis result information, it is determined that the combination information of “major category” and “minor category” in the row exists, and matches with “major category” and “minor category”. Memorize the expression. For example, the morphological analysis result of the news ID “191” displays the row for displaying the column name of the evaluation word dictionary in FIG. 3 (that is, “major category”, “minor category”, “headline expression”, “corresponding expression”). The corresponding expression “[^ <>] + <noun-proper noun-organization>” in the first line excluding “To” line matches “/ A company <noun-proper noun-organization>”, and the major classification “object” The subcategory “company” heading expression “organization name” corresponding expression “/ A company <noun-proper noun-organization>” and news ID “191” are stored. In the news ID “188”, there is only one matching expression, but there are cases where the expression matches in a plurality of lines of the evaluation word dictionary and a plurality of sets of information are stored.

  Based on whether the identity determination criteria in the storage unit 11 are satisfied, the identity determination unit 15 determines that the latest ID-added news information in the storage unit 11 and the ID-added news information written in the past are the same news. It is determined whether or not there is (steps S4 to S9).

  In step S4, the identity determination unit 15 extracts the ID-added news information written in the past in the storage unit 11 one by one in order to determine the identity with the latest ID-added news information (step S4). ). Here, the ID-added news information to be extracted is, for example, within a set time such as “the difference in distribution date and time is within one day (24 hours)”.

  In step S5, the identity determination unit 15 confirms the presence / absence of ID-added news information to be determined for identity (step S5). If there is ID-added news information that is subject to identity determination, the ID-added news information is extracted one by one, and the identity determination with the latest ID-added news information is executed (step S6).

  In step S7, the identity determination unit 15 determines whether or not the news source names in the two pieces of ID-added news information match. If not, the determination in steps S8 and S9 is omitted. The process proceeds to S11. For example, if the identity determination target is the news ID “188” shown in FIG. 4 for the latest ID-added news information having the news ID “191”, the identity determination unit 15 The news source name “G newspaper” is determined to match, and the process proceeds to step S8.

  In step S8, the identity determination unit 15 determines whether or not the distribution time difference (time stamp time difference) is within the range considered as the same news in the two pieces of news information with ID. Here, the determination based on the delivery time difference will be supplementarily described.

  News data that is the source of news information is distributed from the news source device to the news distribution site device via the Internet. The news distribution site device checks the distributed news data by the CPU executing the distribution site management program, determines whether or not it is worth posting, and if it is to be posted, Consider title expression (eg, place company name at the beginning or end), and add a time stamp indicating the date and time of news source distribution or posting on your site. Thereby, even for the same news distributed from the same news source, the time stamp and the title of the news may be slightly different depending on the news distribution site. In this case, the time difference between the time stamps is about 5 minutes at the maximum. Therefore, in step S8, whether or not two pieces of ID-added news information are regarded as the same news is determined based on whether or not the two pieces of ID-added news information are within the maximum distribution time difference range. . The above is the supplementary explanation of the determination based on the distribution time difference.

  When the determination result of step S8 is negative, the identity determination unit 15 omits the determination of step S9 and proceeds to step S11. For example, if the identity determination target is the news ID “188” shown in FIG. 4 for the latest ID-added news information having the news ID “191”, the identity determination unit 15 Then, it is determined that the distribution time difference between them (3 minutes) is within the range of the maximum distribution time difference (eg, 5 minutes), and the process proceeds to step S9.

  In step S9, the identity determination unit 15 calculates the similarity based on the independent words and the numerical information in the morphological analysis results of the two pieces of news information with ID, and whether the similarity is higher than the specified value. It is determined whether or not (step S9). If not, the process proceeds to step S11. An independent word refers to a word that can constitute a phrase alone with respect to an attached word. Nouns, pronouns, verbs, adjectives, adjective verbs, adverbs, conjunctions, conjunctions, and emotional verbs correspond to independent words.

  FIG. 13 is a schematic diagram for explaining independent words extracted from the morphological analysis results in the analysis result information including the news ID “188” and the news ID “191”. The independent words extracted from the morphological analysis result of the news ID “191” are “product X <noun-general>”, “part Z <noun-general>”, “improvement <noun-sa-change>”, “last month <noun-adverb”. Possible> "" min <noun-suffix-general> "" production <noun-sa-variant connection> "" electronic control <noun-general> "" adjustment <noun-sa-variant connection> "" Company A <noun-proper noun-organization " > ”.

  The independent words extracted from the morphological analysis result of the news ID “188” as the identity determination target are “Company A <noun-proper noun-organization>” “product X <noun-general>” “part Z <noun-general”. > "Improvement <noun-sa-variant connection>" "Last month <noun-adverb possible>" "Min <noun-suffix-general>" "Production <noun-sa-variant connection>" "Electronic control <noun-general>" Nine types of <noun-sa-variant connection>.

  The independent words extracted from the morphological analysis results of the two news information with IDs are completely the same although the order is different. Two morphological analysis results do not include numerical information. For this reason, the independent word and numerical information in one morphological analysis result and the independent word and numerical information in the other morphological analysis result completely match among two morphological analysis results.

  Therefore, as a result of the determination in step S9, the identity determination unit 15 sets the similarity between independent words and numerical information in the morphological analysis result corresponding to each of the news ID “191” and the news ID “188” to 1.0. It is determined that the similarity is higher than the specified value.

  In addition, since all the determination results in steps S7 to S9 are positive, the identity determination unit 15 receives the ID-added news information including the news ID “188” for the ID-added news information including the news ID “191”. Judge that the news is the same.

  Next, when the same news exists for the latest ID-added news information including the news ID “191”, the importance level calculation unit 16 adds the importance level for each news item to the importance level of the news ID “191”. Is added (step S10).

  In this step S10, the importance level may be added, for example, by setting the initial value of the importance level to 0, and adding, for example, the value “1” each time the presence of the same news is determined. The addition of importance is not limited to addition, but the initial value of importance is set to a value other than 0, and every time it is determined that the same news exists, for example, a predetermined value (however, a value excluding 0 and 1) is multiplied. May be. Moreover, the addition of the importance may be performed every time the determination result of step S9 is obtained, and is executed collectively after the identity determination processing of the identity determination unit 15 is completed for all objects of identity determination. May be.

  On the other hand, if any of the determination results in steps S7 to S9 is negative, the process proceeds to step S11. Specifically, the latest ID-added news information is the news ID “38” in FIG. 4, and the morphological analysis result in step S2 is an example corresponding to the news ID “38” in FIG. I will explain.

  In this case, in step S3, the first row, the second row, and the fourth row of the evaluation word dictionary in FIG. 3 match the morphological analysis result corresponding to the news ID “38”, and the major classification “target "Small category" Company "heading expression" Organization name "corresponding expression" / A company <noun-proprietary noun-organization> ", Large classification" Evaluation "Small classification" Negative "heading expression" Sales decrease "Corresponding expression" Sales <Noun " -General> /> <symbol-closed parenthesis> / Company A <noun-proper noun-organization> /, <symbol-reading> / 1 month <noun-adverbable> / 1 <noun-number> / 5 <noun- <Number> /. <Noun-number> / 8 <noun-number> /% <noun-suffix-classifier> / decrease <noun-suffix-general>], major classification "evaluation" minor classification "negative" heading expression "recall The evaluation result including the corresponding expression “recall <noun-sa change connection>” and the news ID “38” is stored. It is written to the 11.

  Then, in step S4, two pieces of ID-added news information of news ID “3” and news ID “31” in FIG. 4 are extracted. Thereafter, in the identity determination of the news information with ID of the news ID “38” and the news ID “3”, it is determined in step S7 that the news source names are different, and step S11 is executed.

  In step S11, the importance calculation unit 16 calculates the degree of coincidence between the numerical information extracted from the morphological analysis result corresponding to the news ID “38” and the news ID “3” and the independent word. The numerical information may adopt the morphological analysis result as it is. For example, the part of speech “<noun-number>” may be read as a single numerical value, or the part of speech “<noun-number>” followed by the part of speech “<noun-” The word up to the suffix--a classifier> "may be read as a numerical value. Here, the word sequence of part-of-speech “<noun-number>” and the subsequent word “<noun-suffix-classifier>” are collected as numerical information and distinguished from independent words.

  FIG. 14 is a schematic diagram for explaining numerical information and independent words extracted from the news ID “38” and the news for identity determination.

  From the morphological analysis result for the latest ID-added news information including the news ID “38”, one kind of numerical information “15.8% reduction”, independent words “US <noun-proper noun-country>” “product sales <noun” -General> "Company A <Noun-Proper Noun-Organization>" January <Noun-Adverb possible> "Recall <Noun-Sabari Connection>" Problem <Noun-Nay adjective stem> "Sounding <Verb- 7 types of “independence>” are extracted.

  From the morphological analysis results for past ID-added news information including the news ID “3” that is the identity determination target, one type of numerical information “15% reduction <numerical information>”, independent word “Company A <noun-proper noun-” “Organization>” “January <noun-adverb possible>” “rice <noun-proper noun-country>” “product sales <noun-general>” are extracted.

  Here, the numerical information corresponding to the news ID “38” and the news ID “3” are “15.8% reduction” and “15% reduction”, respectively, and the values do not match. In addition, since step S11 is a process of another news, it is not always necessary to match the digits of the significant digits of the numerical information.

  Regarding independent words, there are seven types of independent words extracted from two morphological analysis results, whereas there are four types of independent words extracted from both two morphological analysis results.

  In this case, based on the fact that there are 4 types of matching information among a total of 8 types of information of 1 type of numerical information and 7 types of independent words, the degree of coincidence is calculated as 50% and the degree of similarity is calculated as 0.5.

  The calculation of the similarity is further executed as “the similarity is 0 when the numerical information is included and does not match” or “the similarity between the numerical information and the average of the independence words is the similarity”. May be.

  For example, the importance calculation unit 16 adds the similarity “0.5” corresponding to the news ID “3” to the importance of the news ID “38”. When the importance corresponding to the news ID “3” has already been set, a value obtained by multiplying the importance corresponding to the news ID “3” by the similarity “0.5” is added to the importance of the news ID “38”. You may do that.

  Similarly, in the identity determination of the news information with ID of the news ID “38” and the news ID “31”, it is determined in step S7 that the news source names are different, and step S11 is executed.

  In step S11, the importance level calculation unit 16 calculates the degree of coincidence between the numerical information extracted from the morphological analysis result corresponding to the news ID “38” and the news ID “31”, and the independent word.

  14, from the morphological analysis result corresponding to the news ID “31”, one kind of numerical information “16% reduction <numerical information>”, independent words “US <noun-proper noun-country” ”“ Company A <noun ” -Proper noun-organization "" January <noun-adverb possible> "" sales <noun-sa-variant connection> "" recall <noun-sa-variant connection> "" problem <noun-nai adjective stem> "" sounding <verb- 7 types of “independence>” are extracted.

  Here, the numerical information corresponding to the news ID “38” and the news ID “31” are “15.8% reduction” and “16% reduction”, respectively, and the values do not completely match.

  For independent words, the independent words extracted from the two morphological analysis results completely match. The matching of the numerical information is not limited to being the same with a perfect match, but may be determined whether or not they match in consideration of a difference in significant digits.

  For example, the numerical information “16% decrease” corresponding to the news ID “31” has the first significant digit, and the numerical information “15.8% decrease” corresponding to the news ID “38” has the first significant digit. First place. If the effective number of the numerical information corresponding to the news ID “38” is set to the first place in accordance with the news ID “31”, it is “decreased by 16%”, and the numerical information of the two coincides.

  If the match of the numerical information is determined by combining the digits of significant digits, the morphological analysis result corresponding to the news ID “38” and the news ID “31” completely matches the numerical information / independent words.

  In this case, the degree of coincidence is calculated to be 100% and the degree of similarity is 1.0 based on the fact that there are 8 types of matching information among a total of 8 types of information of 1 type of numerical information and 7 types of independent words. The method of calculating the similarity by combining the significant digits of the numerical information as described above is the same in step S9.

  The importance calculation unit 16 adds the similarity “1.0” corresponding to the news ID “31” to the importance of the news ID “38”. When the importance level corresponding to the news ID “31” has been set, a value obtained by multiplying the importance level corresponding to the news ID “31” by the similarity “1.0” is set as the importance level of the news ID “38”. You may add.

  As described above, when the identity determination with all the identity determination target news is completed and the calculation of the importance level is completed (step S12), the importance level calculation unit 16 determines the news ID in the latest ID-added news information. The evaluation result including the news ID “38” that matches “38” is read from the storage unit 11, and the analysis result including the read evaluation result and the calculated importance is transmitted to the user terminal device 20 through the input / output interface 18. (Step S13).

  FIG. 10 is a schematic diagram showing an example of the analysis result. The analysis result includes, for example, the evaluation result written by the information evaluation unit 14, the distribution time, the news source name, the distribution site name, and the importance calculated by the importance calculation unit 16. The order of information in the analysis result is not limited to this example, and an arbitrary order is applicable.

  In the user terminal device 20, when this analysis result is received by the communication interface 25, the display unit 23 displays the analysis result. Thereby, the user terminal device 20 can present the analysis result of news information to the user of the own device.

  As described above, according to the present embodiment, by adding numerical information, distribution time difference, and news source name to the news information identity determination criterion, it is a more detailed viewpoint than news related to the same topic, that is, whether the news is the same news. It can be determined whether or not. In addition, the company name is determined from the expression in the news title, and the content of the news is classified into “Negative”, “Positive”, “Unstable”, etc., and the information is quickly grasped by the user conducting stock trading. Can help.

  In this embodiment, the analysis result is not limited to the user terminal device 20 but may be modified so as to be transmitted to an algorithm trade engine that makes a stock transaction judgment. In this case, the algorithm trade engine supports the judgment of the stock trade. can do.

  In addition, the input / output interface 18 may distribute the news (title) with the news ID “38” and link information that causes the user terminal device 20 to access the news text together with the transmission of the analysis result. Further, the input / output interface 18 may transmit the analysis result every time the analysis result of the latest ID-added news information is obtained, or may transmit the new analysis result collectively at time intervals set by the user. Good.

  Furthermore, in the present embodiment, the case where the input / output interface 18 transmits the analysis result to the user terminal device 20 has been described. The analysis result may be transmitted to the specified user terminal device 20 of a specific company, and the input / output interface 18 includes the specified evaluation result according to the transmission content specified by the user and is not specified. You may make it transmit to the user terminal device 20 which transmits the analysis result which does not contain an evaluation result.

  Independent word comparison by the identity determination unit 15 compares the morphological analysis results of the word analysis unit 13 as they are, but is not limited thereto, and the verbs / adjectives / adjective verbs in the morphological analysis results are converted into original forms for comparison. It may be modified to add processing such as processing or processing to make a negative final form when returning to the original form when a negative auxiliary verb has continued.

<Second Embodiment>
Next, a second embodiment will be described with reference to the drawings described above.

  The second embodiment is a modification of the first embodiment, and has a configuration in which the terms extracted by the information evaluation unit 14 are taken into account in similarity calculation based on independent words extracted by the identity determination unit 15. ing. Accordingly, the similarity (s) includes the number of terms (n1) in the entire term set extracted by the information evaluation unit 14 from the morphological analysis results of the two titles, and the morphological analysis results of the two titles. For the total number (n3 = n1 + n2) of the number of independent words and numerical information (n2) in the whole set of independent words and numerical information excluding the extracted term from the independent words and numerical information included in The number of terms (m1) extracted from both morphological analysis results of the two titles and the terms extracted from both independent words and numerical information contained in both of the two title morphological analysis results This is the ratio (s = m3 / n3) of the total number (m3 = m1 + m2) of the independent word and the number of numerical information (m2) excluding.

  In addition, in the second embodiment, the evaluation word dictionary shown in FIG. 15 is used.

  Specifically, as the evaluation word dictionary, in addition to the information shown in the example of FIG. 3, a dictionary that associates the formal company name, TSE code, the industry to which it belongs, and the capital group to which it belongs to the expression corresponding to the organization name is used. Yes. In addition, as the evaluation word dictionary, in the dictionary of FIG. 3, “continuous fall” “fall” is “stock price drop”, “US” “USA” is “USA”, “Tokyo Stock Market” “Nikkei Average” is “stock market” ”And a dictionary line for synonym summarization. That is, the “corresponding expression” in the evaluation word dictionary includes the synonym when the term corresponding to the corresponding expression has a synonym.

  Next, the operation of the news information analysis apparatus configured as described above will be described.

  Steps S1 to S2 are executed in the same manner as described above.

  In step S3, the information evaluation unit 14 extracts terms from the analysis result information in the storage unit 11 using the evaluation word dictionary as shown in FIG. Express news analysis results in one line. Specifically, the news ID and news (title) of the analysis result information, and terms extracted from the analysis result information (“target”, “target specific word”, “company name”, “TSE code”, “industry”) , “Evaluation”, “evaluation specific word”, “other headline expression” not yet presented, and “other specific word” corresponding to the other headline expression) are written in the storage unit 11.

  Here, “other headline expressions” are “headline expressions” extracted from the news, which are not presented in “target specific words” or “evaluation specific words”. The “other specific word” is a “specific word” corresponding to this “other headline expression”, that is, a specific character string matching the “corresponding expression” in the news (title).

  In addition, the extraction result of the information evaluation unit 14 has the effect of synonym summarization, such as the company name is aggregated into the official name, and “Nikkei average” and “Tokyo stock market” are aggregated into the “stock market”. ing.

  Steps S4 to S8 are executed in the same manner as described above.

  In step S9 or S11, the identity determination unit 15 is similar based on “numerical information”, “headline expression extracted by the information evaluation unit”, and “independent word extracted from the portion excluding the extraction target of the numerical information / information evaluation unit”. Calculate the degree. In addition, before calculating the similarity, the identity determination unit 15 deletes numerical information corresponding to the TSE code when the company name and the TSE code of the company name appear at the same time. FIG. 17 shows an example of independent word / numerical information extracted from the morphological analysis result in the analysis result information of FIG.

  When the extraction result of the information evaluation unit 14 in FIG. 16 is reflected, the information extracted from the morphological analysis result in the analysis result information having the news ID “183” is 2 of the evaluation words “A machine (stock)” and “stock market”. Species, self-supporting words “small <noun-adjective verb stem>” “continuation <noun-sa-variant connection>” “soft tone <noun-general>” “continuation <noun-sa-variant connection>” “upper <noun-general>” “heavy <Adjective-independence> ”.

  The information extracted from the morphological analysis result in the analysis result information having the news ID “185” is two kinds of evaluation words “A Machine Co., Ltd.” and “stock market”, and the independent words “closeness <noun-general>” “small” <Noun-adjective verb stem> "" Continuous <noun-sa-variant connection> "" Soft tone <noun-general> "" Continuation <noun-sa-variant connection "" "Upper <noun-general>" "Heavy <adjective-independent>" There are seven types.

  The numerical information “<1234>” is deleted because it matches the TSE code of A machine Co., Ltd.

  As a result, 8 kinds out of a total of 9 kinds of evaluation words and independent words appear in both news, and the similarity is calculated as 0.89. By replacing the part extracted by the information evaluation unit 14 with a heading expression registered in the evaluation word dictionary in parts other than numerical information, synonyms expressed in different notations / words are regarded as the same, and the accuracy of identity determination Can be improved.

  Steps S10 and S12 are executed in the same manner as described above.

  In step S <b> 13, the input / output interface 18 can transmit the analysis result to the user terminal device 20 as shown in FIG. 18. For example, when the information evaluation unit 14 assigns formal name information to the company name, it is possible to transmit the analysis result using the formal name. Further, the control unit 17 searches the evaluation word dictionary from the TSE code given to the official name and the industry name, and obtains the TSE code for the company name, the other company name of the industry to which it belongs, and the capital group to which it belongs. The analysis result including the first related information including the company name can also be transmitted from the input / output interface 18. In addition, if the news title has the expression “Major Major” and the pattern match in the last line of the evaluation word dictionary and the major classification “target”, minor classification “industry”, the heading expression “machine” is extracted, the evaluation word dictionary The control unit 17 may search, and the analysis result including the second related information including the company name classified as the industry “machine” obtained as a search result and its TSE code may be transmitted from the input / output interface 18. . Note that both the first and second related information may be included in the analysis result, and at least one may not be included in the analysis result.

  As described above, according to the present embodiment, the accuracy of identity determination can be further improved in addition to the effects of the first embodiment by the configuration in which the corresponding expression in the evaluation word dictionary includes synonyms.

  Although the first and second embodiments do not refer to the use of the evaluation word dictionary, the evaluation word dictionary should be selected and used from among a plurality of evaluation word dictionaries according to the news field and the content that the user is interested in. You can also.

<Third Embodiment>
Next, a third embodiment will be described with reference to the drawings described above.

  The third embodiment is a modification of each of the first and second embodiments, and has a configuration that considers synonyms such as “$ 100 million” and “9.2 billion yen” in numerical information.

  Accordingly, the exchange unit information indicating the exchange ratio between the home currency and the foreign currency is further stored in the storage unit 11.

  Further, in addition to the above-described functions, the identity determination unit 15 has morphological analysis results of two titles in units of currency and units of quantity (for example, kg and pounds, km and yards, international unit systems and yard / pound methods, Or numerical information including those with prefixes in international unit systems such as m and cm), and the numerical information is exchange rate based on exchange rate information, or conversion ratio based on contrast information between units, Before calculating the degree of similarity, the numerical information including the currency unit of the foreign currency and the predetermined unit of quantity corresponds to the currency unit of the home currency and the predetermined unit of quantity. The first numerical information replacement function is provided for replacing the numerical information including other quantity units.

  Next, the operation of the news information analysis apparatus configured as described above will be described.

  Steps S1 to S8 are executed in the same manner as described above.

  In step S9 or S11, the identity determination unit 15 refers to the exchange rate information in the storage unit 11 and reads the amount represented by the numerical information before calculating the similarity.

  For example, as shown in FIG. 19, it is assumed that the identity determination unit 15 extracts an independent word from the morphological analysis result in the analysis result information having the news ID “1722” and the news ID “1736” in FIG. At this time, numerical information “over 100 million dollars” corresponding to the news ID “1722” is extracted, and numerical information “9.2 billion yen” corresponding to the news ID “1736” is extracted. Since these two pieces of numerical information have different currency units, they cannot be compared as they are.

  Therefore, the identity determination unit 15 refers to the exchange rate information of the numerical information “07” extracted from the news title or at the time of news transmission, and after converting “$ 100 million” to “9.2 billion yen”, The numerical information “100 million dollars” is replaced with the numerical information “9.2 billion yen”. Thereby, similarity calculation can be performed in a state where the currency units of numerical information are combined, and the accuracy of identity determination can be further improved.

  Steps S10, S12, and S13 are executed in the same manner as described above.

  As described above, according to the present embodiment, the accuracy of the identity determination is further improved in addition to the effects of the first and second embodiments by the configuration in which the currency unit of the numerical information is matched based on the exchange rate information. Can be improved.

<Fourth Embodiment>
Next, a fourth embodiment will be described with reference to the drawings described above.

  The fourth embodiment is a modification of each of the first to third embodiments, and has a configuration in which synonyms such as the presence or absence of a calculation basis (for example, seasonal adjustment) in numerical information are taken into account.

  Accordingly, the news data and the news information include a news text corresponding to the title.

  Further, in addition to the above-described functions, the identity determination unit 15 uses at least one of the news texts corresponding to the two titles as terms related to calculation grounds in numerical information represented by “seasonal adjustment” (hereinafter referred to as “calculation grounds”). If the morphological analysis results of the two titles contain numerical information of different values in the same unit, before calculating the degree of similarity, A second numerical information replacement function is provided that calculates an adjustment value related to the calculation basis from the larger value of the numerical information in accordance with the calculation procedure, and replaces the larger value with the adjustment value.

  As described above, the basis of calculation is not limited to seasonal adjustment. For example, regarding “red letter”, there may be a case where there is an expression of “business deficit” or “current account deficit” and monetary information related thereto in the text.

  Next, the operation of the news information analysis apparatus configured as described above will be described.

  Steps S1 to S8 are executed in the same manner as described above.

  In step S9 or S11, the identity determination unit 15 refers to the news text and corrects the numerical information based on the calculation basis before calculating the similarity.

  For example, the numerical information corresponding to the news ID “38” and the news ID “49” in FIG. 14 are “15.8%” and “8.7%”, which do not match. That is, the morphological analysis results of the two titles include numerical information of different values in the same unit. In addition, in the news texts corresponding to the two titles, there is a description of “down 8.7% from the same month last year (adjusted based on the calculation basis)” and “down 15.8% from the same month last year”. One of them includes calculation basis terms, and there is a difference in the presence or absence of seasonal adjustment in the calculation basis. Furthermore, “total sales are 98,796 units” agree.

  Accordingly, the identity determination unit 15 calculates the adjustment value “8.7%” from the larger value “15.8%” of the numerical information in accordance with a calculation procedure based on a predetermined calculation basis. The larger value is replaced with the adjustment value. Thereby, the precision of identity determination can be improved further.

  If the numerical information does not match after the numerical information is replaced with the adjustment value, it may be treated as another news. In addition, even if the processing of the third and fourth embodiments is added, the numerical information does not match, but when the matching degree of the evaluation word / independent word is a predetermined value (for example, 0.8) or more, it is corrected with the same topic It may be determined that the information is important news to be distributed, and the importance calculation unit 16 may add a prescribed numerical value to the importance.

  Steps S10, S12, and S13 are executed in the same manner as described above.

  As described above, according to the present embodiment, in addition to the effects of the first to third embodiments, the identity determination is performed by the configuration in which the seasonal adjustment value of the numerical information is matched based on the calculation basis term of the news body. The accuracy can be further improved.

<Fifth Embodiment>
Next, a fifth embodiment will be described with reference to the drawings described above.

  The fifth embodiment is a modification of the first to fourth embodiments, and is configured to correct the importance when the frequency of extracted terms is high.

  Accordingly, for example, the control unit 17 extracts the extracted terms as shown in FIG. 20 based on the extracted terms included in the evaluation results in the storage unit 11 and the distribution time in the news information with ID. A statistical information writing function for writing statistical information including the term and the appearance frequency of the term into the storage unit 11 is further provided.

  Further, the importance calculation procedure in the storage unit 11 is the importance calculated by the calculation procedure when calculating the importance of the ID-added news information including a word whose appearance frequency indicated by the statistical information exceeds the reference frequency. And a procedure for correcting the importance by adding an addition value based on a word exceeding the reference frequency to the degree. In addition, as a technique that can determine whether or not the appearance frequency indicated by the statistical information exceeds the reference frequency, for example, in recent years, a “rapidly rising word” has been published on a search site or the like. Techniques for determining are known.

  Next, the operation of the news information analysis apparatus configured as described above will be described.

  Steps S1 to S3 are executed in the same manner as described above.

  After step S3, the control unit 17 determines the extracted term and the appearance frequency of the term based on the extracted term included in the evaluation result in the storage unit 11 and the distribution time in the news information with ID. Is written in the storage unit 11.

  Steps S4 to S11 are executed in the same manner as described above.

  In step S12, the importance calculation unit 16 adds information according to the importance of the words included in the news title when determining the importance of the latest ID-added news information.

  The importance level calculation unit 16 refers to the statistical information in the storage unit 11 based on the importance level calculation procedure in the storage unit 11, and the ID-added news information includes words whose appearance frequency indicated by the statistical information exceeds the reference frequency. In this case, the importance level is corrected by adding an addition value based on a word exceeding the reference frequency to the importance level already calculated for the ID-added news information.

  Step S13 is executed as described above.

  As described above, according to the present embodiment, the control unit 17 has a statistical information writing function, and the importance calculation procedure in the storage unit 11 is configured to correct the importance based on words exceeding the reference frequency. In addition to the effects of the first to fourth embodiments, it is possible to notify the user of an analysis result in which news information including words with high appearance frequency is added with high importance.

  In addition, when the news information with ID includes a word exceeding the reference frequency, the importance calculating unit 16 replaces the process of correcting the importance with information notifying that the word includes a word exceeding the reference frequency (rapidly rising word). It may be included in the analysis results. By notifying the user of the presence of a rapidly rising word, the user can easily recognize the occurrence of a new event that is likely to give a new fluctuation to the stock price.

<Sixth Embodiment>
Next, a sixth embodiment will be described with reference to the drawings described above.

  The sixth embodiment is a modification of each of the first to fifth embodiments, and is configured to correct the importance based on the reliability corresponding to each of the distribution site name and the news source name.

  Accordingly, in addition to the stored contents described above, the storage unit 11 stores the distribution site name or news source name and the reliability for the distribution site name or the news source name in association with each other as shown in FIG. ing. As the reliability, for example, “1” is set when the reliability of the news data or news information to be distributed is high, and according to the reliability when the reliability of the news data to be distributed is low, for example. Set a value less than 1.

  In addition, the importance level calculation procedure in the storage unit 11 calculates the importance level of news information including the distribution site name and the news source name to the distribution site name with respect to the already calculated importance level. And a procedure for correcting the importance by multiplying the corresponding reliability by the reliability corresponding to the news source name.

  Next, the operation of the news information analysis apparatus configured as described above will be described.

  Steps S1 to S11 are executed in the same manner as described above.

  In step S12, the importance calculation unit 16 performs adjustment based on the reliability of the identity determination target news distribution site and the news source when calculating the importance based on the similarity.

  The importance level calculation unit 16 has, for example, a similarity with the past ID-added news information of 0.8, the distribution site name corresponding to the latest ID-added news information is F news, and the news source name is I newspaper. In this case, similarity 0.8 × distribution site reliability 0.9 × news source reliability 0.7 = 0.504 is added to the importance for the latest ID-added news information. The calculation of the reliability of the distribution site and the news source is not limited to multiplication, and another calculation such as addition may be used.

  Step S13 is executed as described above.

  As described above, according to this embodiment, in addition to the effects of the first to fifth embodiments, the importance is corrected based on the reliability corresponding to each of the distribution site name and the news source name. Considering the reliability of distribution sites and news sources and reducing the impact of news with low reliability, it is possible to provide the user with analysis results including appropriate importance.

  According to at least one embodiment described above, news source names included in two pieces of news information match each other, the difference in delivery date and time is smaller than a reference value, and the similarity is higher than a specified value. The latest ID-added news information and the ID-added news information written in the past are distributed based on whether or not the identity determination criteria including While performing identity determination with respect to a plurality of news information, the accuracy of identity determination can be improved.

  Supplementally, not only the similarity of independent words and numerical information but also the match of news source names and the small difference in distribution time are used as criteria for determining the identity of news, for example, “Product X recall of company A” And “Additional recall of product Y at Company A” (for example, due to a mismatch in news source names or a large difference in distribution time), and providing new information useful for the judgment of stock trading to the user of the user terminal device 20 Can do.

  In addition, according to at least one embodiment, the similarity is calculated after rounding the significant digits of the numerical information by rounding, so that the numerical information of the distributed news and the numerical information of other information are identical. Therefore, it is possible to easily cope with information correction and to further improve the accuracy of identity determination.

  Furthermore, according to at least one embodiment, unlike the conventional case, the user terminal device is configured to transmit the analysis result including the importance based on the number of the same news after the identity determination to the user terminal device 20. In 20, many news items having the same identity are not displayed, and the user can easily grasp the whole image.

  In addition, the method described in each said embodiment does not limit an object to Internet delivery news. For example, it is possible to target other information distributed on the Internet or electronic documents existing outside the Internet.

  In addition, the method described in each embodiment includes, as programs that can be executed by a computer, a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), an optical disk (CD-ROM, DVD, etc.), a magneto-optical disk (MO). ), And can be distributed in a storage medium such as a semiconductor memory.

  In addition, as long as the storage medium can store a program and can be read by a computer, the storage format may be any form.

  In addition, an OS (operating system) running on a computer based on an instruction of a program installed in the computer from a storage medium, MW (middleware) such as database management software, network software, and the like realize the above-described embodiment. A part of each process may be executed.

  Furthermore, the storage medium in each embodiment is not limited to a medium independent of a computer, but also includes a storage medium in which a program transmitted via a LAN, the Internet, or the like is downloaded and stored or temporarily stored.

  Further, the number of storage media is not limited to one, and the case where the processing in each of the above embodiments is executed from a plurality of media is also included in the storage media in the present invention, and the media configuration may be any configuration.

  The computer in each embodiment executes each process in each of the above embodiments based on a program stored in a storage medium, and a single device such as a personal computer or a plurality of devices are connected to a network. Any configuration of the system or the like may be used.

  In addition, the computer in each embodiment is not limited to a personal computer, and includes an arithmetic processing device, a microcomputer, and the like included in an information processing device, and is a generic term for devices and devices that can realize the functions of the present invention by a program. Yes.

  In addition, although some embodiment of this invention was described, these embodiment is shown as an example and is not intending limiting the range of invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  DESCRIPTION OF SYMBOLS 10 ... News information analyzer, 11 ... Memory | storage part, 12 ... News receiving part, 13 ... Word analysis part, 14 ... Information evaluation part, 15 ... Identity determination part, 16 ... Importance calculation part, 17 ... Control part, 18 DESCRIPTION OF SYMBOLS Input / output interface, 19 Bus, 20 User terminal device, 21 Terminal control device, 22 Terminal storage unit, 23 Display unit, 24 Input unit, 25 Communication interface

Claims (7)

When news data including a title described in Japanese is distributed from a news source device of a news source, a distribution date and time indicating the date and time of distribution or a date and time posted on a news distribution site, and the news distribution site are indicated. A news distribution site device of the news distribution site that transmits news information including a distribution site name, a news source name indicating the news source, and a title that is the same as the title of the news data or a title in which a word of the title is replaced; ,
A news information analysis apparatus comprising a memory that can be individually communicated with both devices with a user terminal device to which the analysis result of the news information is transmitted,
A corresponding expression indicating a regular expression of a word analysis result for a term including one or more words that may be included in the title of the news information, a heading expression corresponding to the corresponding expression, and a meaning of the heading expression An evaluation word dictionary storage means for storing an evaluation word dictionary in which a small classification and a large classification indicating the meaning of the small classification are associated;
An identity determination criterion indicating a criterion for determining whether or not any two pieces of news information in the transmitted news information are the same news, and news source names included in the two pieces of news information are mutually A match, a difference in distribution time indicating a difference in distribution date and time included in the two news information items is smaller than a reference value, and an independent word in the word analysis result of two titles in the two news information items And the similarity criterion calculated from the numerical information is higher than a prescribed value, and the similarity is calculated after rounding the significant digits of the numerical information by rounding off. Identity criteria storage means;
Important showing a procedure for calculating the importance of the news information based on the number of news information determined to be the same news and the similarity calculated from the word analysis result when determined to be no Importance calculation procedure storage means for storing the degree calculation procedure;
When receiving news information from the news distribution site device, news receiving means for adding a news ID to the news information and writing the obtained ID-added news information in the memory;
A word that analyzes the title included in the news information with ID in the memory, adds the obtained word analysis result to the news ID and title of the news information with ID, and writes the obtained analysis result information to the memory Analysis means;
The term corresponding to the corresponding expression is extracted from the word analysis result in the analysis result information in the memory, and the major classification and the minor classification related to the corresponding expression are extracted from the evaluation word dictionary and extracted. An evaluation result writing means for writing an evaluation result including a term, a major classification, and a minor classification in association with a news ID matching the news ID of the analysis result information,
Whether or not the latest ID-added news information in the memory and the ID-added news information written in the past are the same news based on whether or not the identity determination criterion in the identity determination criterion storage means is satisfied. Identity determining means for determining
Importance calculating means for calculating the importance of the latest ID-added news information based on the importance calculating procedure in the importance calculating procedure storage means and the determination result by the identity determining means;
An evaluation result associated with a news ID that matches a news ID in the latest ID-added news information is read from the memory, and an analysis result including the read evaluation result and the calculated importance is displayed on the user terminal device. Analysis result transmission means for transmitting to
A news information analyzing apparatus comprising: In the news information analysis device according to claim 1,
The similarity is included in both the word analysis results of the two titles with respect to the number of independent words and numerical information in the whole set of independent words and numerical information included in the word analysis results of the two titles. A news information analyzing apparatus characterized in that it is a ratio occupied by the number of independent words and numerical information. In the news information analysis device according to claim 1,
The corresponding expression includes the synonym when the term corresponding to the corresponding expression has a synonym,
The similarity is the number of the terms in the entire set of terms extracted by the evaluation result writing means from the word analysis results of the two titles, the independent words included in the word analysis results of the two titles, and Extracted from both the word analysis results of the two titles for the number of independent words excluding the extracted term from the numerical information and the total number of independent words and numerical information in the entire set of numerical information. And the number of independent words and numerical information excluding terms extracted from both words and numerical information included in both word analysis results of the two titles. A news information analyzer characterized by a ratio. In the news information analysis device according to any one of claims 1 to 3,
A ratio information storage means for storing a conversion ratio between the home currency and the foreign currency or a conversion unit of a predetermined quantity unit and a quantity unit corresponding to the predetermined quantity unit;
The identity determination means includes
If the word analysis results of the two titles include numerical information including currency units or quantity units, and the numerical information can be regarded as synonymous based on the exchange ratio or the conversion ratio, before calculating the similarity In addition, among the numerical information, the numerical information including the currency unit of the foreign currency and the predetermined amount unit is replaced with the numerical information including the currency unit of the home currency and another amount unit corresponding to the predetermined amount unit. Numerical information replacement means,
A news information analyzing apparatus comprising: In the news information analysis device according to any one of claims 1 to 4,
The news data and the news information include a news body corresponding to the title,
The identity determination means includes
When at least one of the news texts corresponding to the two titles includes a calculation basis term and the word analysis results of the two titles include numerical information of different values in the same unit, the similarity Before calculating the adjustment value, the adjustment value for the calculation basis is calculated from the larger value of the numerical information in accordance with the calculation procedure based on the predetermined calculation basis, and the larger value is adjusted. Second numerical information replacement means for replacing with a value;
A news information analyzing apparatus comprising: In the news information analysis device according to any one of claims 1 to 5,
Statistics that write statistical information including the extracted terms and the frequency of appearance of the terms to the memory based on the extracted terms included in the evaluation results in the memory and the distribution time in the news information with ID An information writing means,
In the importance calculation procedure, when calculating the importance of ID-added news information including a word whose appearance frequency indicated by the statistical information exceeds a reference frequency, the importance is calculated with respect to the importance calculated by the calculation procedure. A news information analysis apparatus comprising: a procedure for adding an addition value based on a word exceeding the frequency to correct the importance. The news information analysis apparatus according to any one of claims 1 to 6,
A reliability storage means for storing the distribution site name or the news source name and the reliability corresponding to the distribution site name or the news source name in association with each other;
The importance calculation procedure corresponds to the distribution site name with respect to the importance calculated by the calculating procedure when calculating the importance of news information including the distribution site name and the news source name. A news information analyzing apparatus comprising: a procedure for correcting the importance by multiplying the reliability and the reliability corresponding to the news source name.

Images