JP4919386B2 - Information extraction / display device - Google Patents

Description

The present invention relates to an information extraction and display technology, especially relates to the information extraction and display equipment to extract and view trends information from the article group.

  As a conventional technique for extracting trend information, for example, Non-Patent Document 1 below uses the inter-sentence relationship to determine whether the inter-sentence relationship is transition or update, and extracts the trend information using the information. It describes about technology.

Here, trend information refers to information that focuses on the quantity of a certain item and summarizes its temporal changes. Examples include a change in the number of homeruns of a person and a change in the cabinet support rate. To extract trend information, it is not enough to extract an expression of a certain item and its quantity. It is also necessary to extract a time expression corresponding to the quantity at the same time.
Namba, Kokusei, Fukushima, Aizawa, Okumura: “Extraction and visualization of trend information considering inter-document relations”, IPSJ SIG, 2005-NL-168, pp.67-74 (2005) .

  However, the above prior art does not automatically extract the unit representation of the numerical information or automatically extract the item corresponding to the numerical information. Therefore, depending on the prior art, main unit expressions, time expressions, and item expressions are automatically extracted from articles related to a certain field, trend information is extracted using them, and the extracted trend information is graphed. Is difficult to display.

The present invention shows the above-mentioned solution to the problems of the prior art, and to provide information extraction and display equipment to extract and view trends information automatically from article group associated with a field.

  In order to solve the above problems, the present invention provides an information extraction / display device that extracts and displays trend information, and extracts a main unit expression in the article group from an article group related to a certain field. An extraction unit is provided.

  In addition, the present invention is an information extraction / display device that extracts and displays trend information, and is extracted by a main expression extraction unit that extracts a main expression from an article group related to a certain field, and the main expression extraction unit. A trend information pair extracting means for extracting a trend information pair from the article group based on the main expression, and a display means for displaying the trend information pair extracted by the trend information pair extracting means. .

  In the information extraction / display apparatus according to the present invention, the main expression extraction unit includes a main unit expression extraction unit that extracts a main unit expression from the article group, and a main time expression that extracts a main time expression from the article group. It comprises time expression extraction means and main item expression extraction means for extracting a main item expression from the article group.

  In the information extraction / display apparatus according to the present invention, the trend information pair extraction unit extracts the trend information pair by using a machine learning technique.

  The present invention also relates to an information extraction / display device that extracts and displays trend information, and extracts trend information pairs from a group of articles related to a certain field based on the inputted main expression. And a display means for displaying the trend information pair extracted by the trend information pair extraction means, wherein the trend information pair extraction means extracts the trend information pair using a machine learning technique. Features.

  In the information extraction / display apparatus according to the present invention, the main expression extraction unit extracts a plurality of main expressions, and the display unit extracts the trend information extraction unit based on the extracted main expressions. The main trend information pairs are extracted from the plurality of types of trend information pairs, and the extracted main trend information pairs are displayed.

  Further, the present invention is the information extraction / display apparatus, wherein the trend information pair extraction unit further includes the article based on a selected main expression among the main expressions extracted by the main expression extraction unit. It is characterized by extracting trend information pairs from groups.

  Further, the present invention provides a keyword input unit for inputting a keyword and an article group extraction unit for extracting an article group related to the input keyword from bibliographic data stored in a storage unit in the information extraction / display apparatus. The main expression extracting means extracts the main expression from the article group extracted by the article group extracting means.

  Further, the present invention is characterized in that, in the information extraction / display apparatus, the display means displays the trend information pairs extracted by the trend information pair extraction means in a graph.

  In the information extraction / display apparatus according to the present invention, the display unit extracts a sentence including the trend information pair extracted by the trend information pair extraction unit from the article group, and the extracted sentence includes The trend information pair is highlighted.

  The present invention also relates to an information extraction / display device that extracts and displays trend information, and extracts trend information pairs from a group of articles related to a certain field based on the inputted main expression. Means and display means for displaying the trend information pair extracted by the trend information pair extraction means, wherein the display means displays a sentence including the trend information pair extracted by the trend information pair extraction means as the article group. And the trend information pair is highlighted in the extracted sentence.

  Further, the present invention is an information extraction / display method for extracting and displaying trend information, characterized by comprising a step of extracting a main unit expression in the article group from an article group related to a certain field. .

  Further, the present invention is a program that is executed by a computer included in an information extraction / display device that extracts and displays trend information, wherein the computer is configured to represent a main unit in the article group from an article group related to a certain field. It is characterized in that a process of extracting the above is executed.

  According to the information extraction / display apparatus, information extraction / display method, and information extraction / display program of the present invention, when an article group related to a certain field is given, trend information is automatically extracted from the article group and displayed. It becomes possible.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing an example of a system configuration of the present invention. The information extraction / display device 1 is a processing device that extracts and displays trend information from articles related to a certain field. In the embodiment of the present invention, the information extraction / display device 1 may be a processing device that extracts or displays trend information.

  The information extraction / display apparatus 1 includes a main expression extraction unit 11, a trend information pair extraction unit 12, a main trend information pair extraction / display unit 13, and a related article database (DB) 14.

  The main expression extraction unit 11 extracts a main expression from an article group related to a certain field accumulated in a related article DB 14 described later. The main expressions to be extracted are, for example, unit expressions, time expressions, and item expressions. The main expression is used when a trend information pair is extracted by a trend information pair extraction unit 12 to be described later. When extracting a main expression, for example, a corresponding expression that appears uniformly and frequently in the entire target article group is extracted.

  The main expression extraction unit 11 includes a main unit expression extraction unit 111, a main time expression extraction unit 112, and a main item expression extraction unit 113.

  The main unit expression extraction unit 111 extracts unit expressions necessary for extracting and organizing trend information. For example, in the case of an article of home run competition, “books” such as “50” and “50” are extracted as unit expressions. The main time expression extraction unit 112 extracts a time expression necessary for extracting and organizing trend information. For example, time expressions such as day, month, and year are extracted. The main item expression extraction unit 113 extracts item expressions necessary for extracting and organizing trend information. For example, in the case of an article on a home run competition, expressions that are subject to trend surveys such as “Mugwire” and “Sosa” are extracted as item expressions.

  The trend information pair extraction unit 12 identifies a part where the main expressions extracted by the main expression extraction unit 11 appear at the same time in the target article group, and the main unit expression described in the part and the main unit expression A pair of a numerical expression composed of numerical values corresponding to, a main time expression, and a main item expression is extracted as a trend information pair. Here, the numerical value corresponding to the main unit expression corresponds to, for example, a numerical value appearing in the article adjacent to the main unit expression. That is, for the main unit expression, the numerical value related to it is extracted at the same time, and the numerical value and the main unit expression are extracted as a numerical expression. For example, in the case of an article of home run competition, information pairs “item expression: maguire”, “time expression: 11th”, and “numerical information: No. 47” are extracted as trend information pairs.

  The main trend information pair extraction / display unit 13 organizes the trend information pairs extracted by the trend information pair extraction unit 12 in the target article group, and presents the trend information as a graph or highlighted text. For example, in the case of an article of home run competition, the trend information pair regarding “Mugwire” taken out by the trend information pair extraction unit 12 is displayed in a graph. For example, the time axis is taken on the horizontal axis, and the number of home runs is taken on the vertical axis, which are displayed as a graph. For example, the main trend information pair extraction / display unit 13 extracts a sentence including the trend information pair from the article group in the related article DB 14, and highlights the trend information pair in the extracted sentence.

  When there are a plurality of main expressions extracted by the main expression extracting unit 11, the main trend information pair extracting / displaying unit 13 performs a main operation from a plurality of types of trend information pairs extracted by the trend information pair extracting unit 12 based on each main expression. A configuration may be adopted in which, after extracting the trend information pairs, the extracted major trend information pairs are graphed or highlighted.

  The related article DB 14 stores an article group related to a certain field. In the embodiment of the present invention, the related article DB 14 is omitted, and the information extraction / display apparatus 1 extracts the main expression based on the inputted article group related to a certain field, and extracts the trend information. A configuration may be adopted.

  In the embodiment of the present invention, the information extraction / display apparatus 1 further includes a keyword input unit (not shown) for inputting a keyword, and an article group related to the input keyword as storage means (shown). May be provided with an article group extraction unit (not shown) that is extracted from the bibliographic data and stored in the related article DB 14. The bibliographic data corresponds to, for example, a large-scale corpus. The article group extraction unit may input the extracted article group to the main expression extraction unit 11, and the main expression extraction unit 11 may extract the main expression from the input article group. In the embodiment of the present invention, the input keyword itself may be used as the main item expression.

Below, the detailed example of each component of the information extraction / display apparatus 1 which concerns on embodiment of this invention is demonstrated.
(Main Expression Extraction Unit 11)
The main expression extraction unit 11 extracts main expressions necessary for extracting and organizing trend information. The following are extracted as main expressions.

Unit Representation Time Representation Item Representation For example, ChaSen (see Reference (1) below) is used to extract each representation.

Reference (1): Y. Matsumoto, A. Kitauchi, T. Yamashita, Y. Hirano, H. Matsuda and M. Asahara: Japanese morphological analysis system ChaSen version 2.0 manual 2nd edition ”(1999).
In the output of ChaSen, each expression is extracted using part of speech information. The unit representation takes out a noun series that connects to the front or back of the numerical value. As the time expression, for example, an expression including expressions related to time (eg, “year”, “month”, “day”, “hour”, “second”) among expressions obtained as unit expressions is used as a time expression. For the item expression, for example, a noun series is taken out.

  However, in each of the above processes, for example, an unknown word is treated as a noun. In the case of consecutive nouns connected backward, if a noun other than the suffix comes after the suffix noun in the part of speech subcategory, for example, the suffix noun is taken out. Further, for example, a typical error of ChaSen or an unwieldy output is automatically corrected by a program. For example, an error output using an unknown word su or na as a verb or a particle is set as an unknown word. When outputting a numerical value and a unit such as “January” as a single noun, the unit expression is decomposed into “1” and “month” for easy extraction.

  In addition to the method using ChaSen described above, a unit expression, a time expression, and an item expression can be handled as specific expressions, and a method of extracting a main expression using a specific expression extraction technique described below can be adopted.

  A specific expression is a linguistic expression that means a specific thing / quantity, such as a numerical name such as a name, a place name, or an organization name, and a numerical expression such as a monetary amount. This is a technology for automatic extraction by a computer. For example, if a specific expression is extracted for a sentence “the Japanese prime minister is Junichiro Koizumi”, the specific expressions “Japan” and “Junichiro Koizumi” are extracted as place names and personal names.

Hereinafter, an example of a general technique for extracting a specific expression will be described.
(1) A method using machine learning There is a method of extracting a specific expression using machine learning (for example, see the following reference (2)).

  Reference (2): Masayuki Asahara, Yuji Matsumoto, Use of Redundant Morphological Analysis in Japanese Named Expression Extraction Information Processing Society of Japan Natural Language Processing Study Group NL153-7 2002

First, for example, the sentence “Japan's prime minister is Mr. Koizumi” is divided into each character, and the correct characters such as B-LOCATION and I-LOCATION are assigned to the divided characters as follows. Set the correct answer. The first column below is each divided character, and the correct tag of each character is the second column.
Sun B-LOCATION
I-LOCATION
O
Neck O
Phase O
Is O
Small B-PERSON
Izumi I-PERSON
O
N
At O
O
. O
In the above, B-? ? ? Is a tag that signifies the start of the type of proper expression below the hyphen. For example, B-LOCATION means the beginning of a unique expression called place name, and B-PERSON means the beginning of a unique expression called person name. I-? ? ? Is a tag that means something other than the beginning of the type of proper expression below the hyphen, and O is something else. Therefore, for example, the character “day” is a character that corresponds to the beginning of the unique name “place name”, and the character “book” is the unique name “place name”.

  In this way, the correct answer of each character is set, learned from such data, this correct answer is estimated with new data, and from this correct answer tag, the beginning of each proper expression and how far it is. Is recognized and the proper expression is estimated.

When learning from the correct data set for each character, the system uses various information in the form of features. For example,
Sun B-LOCATION
Part of
Japan-B Noun-B
Such information is used. Japan-B means the beginning of the word Japan, and noun-B means the beginning of the noun. For recognition of words and parts of speech, for example, morphological analysis by ChaSen described above is used. If ChaSen is used, the input Japanese can be divided into words. For example, ChaSen divides a Japanese sentence and estimates the part of speech of each word as described above. For example, if “go to school” is entered, the following results can be obtained.

School Gacco School Noun-General To He To particle-Case particle-General Go Iku Go Verb-independence
In this way, each line is divided so that one word is included, and reading and part-of-speech information are given to each word.

  For example, in the above reference (2), as features, characters constituting the input sentence itself (for example, “small” character), character type (for example, hiragana, katakana, etc.), part of speech information, Tag information (for example, “B-PERSON” or the like) is used.

  Learning using these features. Investigate what features appear in the characters that estimate the tag and the surrounding characters, learn what features are likely to appear when the features appear, and use the learning results to create new data Perform tag estimation. For machine learning, for example, a support vector machine is used.

  In addition to the above-described method, there are various methods for extracting the proper expression. For example, there is a technique for extracting a specific expression using a maximum entropy model and a rewrite rule (see reference (3)).

  Reference (3): Kiyotaka Uchimoto, Maoi, Maki Murata, Hiromi Osaku, Masao Uchiyama, Hitoshi Isahara, Named Expression Extraction Based on Maximum Entropy Model and Rewriting Rules, Journal of the Language Processing Society, Vol.7, No.2 , 2000

  Further, for example, the following reference (4) describes a technique for extracting Japanese proper expressions using a support vector machine.

Reference (4): Hiroyasu Yamada, Taku Kudo, Yuji Matsumoto, Extracting Japanese Named Expressions Using Support Vector Machine, Journal of Information Processing Society of Japan, Vol.43, No.1 ", 2002
(2) A method using a created rule There is also a method of manually creating a rule to extract a specific expression.

For example,
Noun + “san” is the name of the person + “prime” is the name of the person + “town” is the name of the place + “city” is the place of name.

  Although the above-described specific expression extraction technique has been described by taking the case of extracting a person name or a place name as an example, in the embodiment of the present invention, unit expression, time expression, and item expression are treated as specific expressions, respectively You may take the structure which extracts a unit expression, a time expression, and item expression using a technique.

  Next, the main unit expressions, time expressions, and item expressions that play the main role in the articles in the field that we are dealing with are extracted. For example, a corresponding expression that appears uniformly and frequently in the entire target article group is extracted as a main expression.

Specifically, for the extraction of the main expression, the score (score) values as shown in the following formulas (1) to (3) are used, and the higher the score value, the more the main expression. Judgment and extract main expressions.
(1) Okapi's TF term equation

(2) Total frequency

(3) Total number of appearing articles

Where i is the article number, Docs is the set of article numbers, TF _i is the number of appearances of the expression in article i, l _i is the length of article i, and Δ is the average length of articles in article group Docs. means. The expression of the TF term of Okapi has the effect of increasing the score of an expression that appears uniformly in a plurality of articles and has a high frequency. The length of the article is, for example, the number of words or characters included in the article.

For the items representation, a long string so that it can fetch give priority to, without the number of occurrences of the expression of the TF _i article i, the string of the number of occurrences and its representation of the representation of the article i The long product method was also used.

  In the embodiment of the present invention, the value obtained by multiplying the value of equation (1) by IDF, that is, log N / DF, the value of equation (2) by the IDF, and the value of equation (3) by the above A value obtained by multiplying IDF may be used as the value of each score. Here, N means the total number of articles in a large-scale corpus (not shown), and DF means the number of articles in which the expression appears in the large-scale corpus, for example.

  In the embodiment of the present invention, the main expression extraction unit 11 extracts, for example, an expression having the highest calculated score value as the main expression. For example, the main expression extraction unit 11 may extract an expression having a calculated score value equal to or greater than a predetermined threshold as the main expression. In addition, the main expression extraction unit 11 may extract a predetermined number of expressions as the main expression from the one with the high calculated score value, for example. In addition, the main expression extraction unit 11 may adopt a configuration in which the extracted expressions are sorted and output in descending order or ascending order with respect to the score values, for example.

  In the embodiment of the present invention, for example, the main unit expression extracting unit 111 includes unit expression extracting means 200, score value calculating means 201, and main unit expression extracting means 202 as shown in FIG. The unit expression extraction unit 200 extracts a unit expression from the article group in the related article DB 14. The score value calculation unit 201 calculates a score value for the extracted unit representation. The main unit expression extraction unit 202 extracts the main unit expression based on the calculated score value.

  Further, for example, the main time expression extracting unit 112 includes a time expression extracting unit 300, a score value calculating unit 301, and a main time expression extracting unit 302 as shown in FIG. The time expression extraction unit 300 extracts a time expression from an article group in the related article DB 14. The score value calculation unit 301 calculates a score value for the extracted time expression. The main time expression extraction unit 302 extracts a main time expression based on the calculated score value.

For example, the main item expression extracting unit 113 includes an item expression extracting unit 400, a score value calculating unit 401, and a main item expression extracting unit 402 as shown in FIG. The item expression extraction unit 400 extracts item expressions from the article group in the related article DB 14. The score value calculation unit 401 calculates a score value for the extracted item expression. The main item expression extraction unit 402 extracts the main item expression based on the calculated score value.
(Trend information pair extraction unit 12)
The trend information pair extraction unit 12 specifies, for example, a part where the expressions extracted by the main expression extraction unit 11 appear simultaneously in the target article group, and based on the main expression pair described in the part, the trend information pair extraction unit 12 Extract information pairs. That is, the trend information pair extraction unit 12 includes, for example, a main unit expression and a main unit expression described in a portion where the main unit expression, the main time expression, and the main item expression appear simultaneously in the target article group. It is a trend information pair extracting means for extracting a pair of a numeric expression composed of corresponding numeric values, a main time expression, and a main item expression as a trend information pair.

  In the embodiment of the present invention, for example, a special symbol indicating a punctuation mark, a line feed, or a document break is used as a break, and a unit expression, a time expression, and an item expression appear at the same time without intervening these. A place. For example, for each article, there is one trend information pair, and only the trend information pair that appears first in the article is extracted. Further, a combination of a unit expression and a numerical value connected to the unit expression is taken out as a numerical expression.

  In the embodiment of the present invention, since a plurality of trend information pairs may be described in one article, the trend information pair extraction unit 12 may adopt a configuration for extracting a plurality of trend information pairs. Further, in the embodiment of the present invention, the trend information pair extraction unit 12 places importance on the information of the part where the unit expression, the time expression, and the item expression appear more closely, and is described in the part. A configuration may be adopted in which trend information pairs are extracted based on the main expression pairs. For example, a trend information pair may be extracted based on a main expression pair described in a place where the number of characters or the number of words between the unit expression, time expression, and item expression is equal to or less than a predetermined threshold.

  In other words, the trend information pair extraction unit 12 describes, for example, the main unit expression and the main item described in the portion where the main unit expression, the main time expression, and the main item expression appear more closely in the target article group. It is a trend information pair extraction means for extracting a pair of a numeric expression consisting of a numerical value corresponding to a unit expression, a main time expression, and a main item expression as a trend information pair.

  In the embodiment of the present invention, as described later, a configuration may be adopted in which trend information pairs are extracted using a machine learning method.

In the embodiment of the present invention, when the main expression extraction unit 11 extracts a plurality of main item expressions, the trend information pair extraction unit 12 may, for example, include the plurality of the plurality of main item expressions in one article in the target article group. In addition to the condition that the main item expression appears at the same time, a configuration may be adopted in which trend information pairs are extracted.
(Main trend information pair extraction / display unit 13)
The main trend information pair extraction / display unit 13 organizes the trend information pairs extracted by the trend information pair extraction unit 12 in the target article group and presents them in a graph or highlighted text. Create a graph with trend information pair time expression on the horizontal axis and numerical expression on the vertical axis. Moreover, the sentence which extracted the trend information pair is extracted from the article group in the related article DB 14, and the trend information pair is highlighted in the sentence.

  In the embodiment of the present invention, when there are a plurality of trend information pairs in the extracted sentence, for example, the trend information pair that appears first is represented by the quantity expression, time, and time included in the trend information pair. It may be displayed separately from expressions, quantity expressions other than item expressions, time expressions, and item expressions. For example, for example, the first trend information pair that appears is a double line, and the quantity expression, time expression, and item expression other than the item expression included in the trend information pair are highlighted with a single line. You may make it do.

  Further, in the embodiment of the present invention, the main trend information pair extraction / display unit 13 is configured such that when the main expression extraction unit 11 extracts a plurality of main unit expressions, main time expressions, and main item expressions, the plurality of expressions. In the plural types of trend information pairs extracted by the trend information pair extraction unit 12 based on the data for all combinations of the above, the more extracted trend information pairs are determined as useful trend information, and the trend information pairs are Main trend information pair. For example, the most extracted trend information pair is extracted as main trend information. Then, the main trend information pair extraction / display unit 13 graphs or highlights the extracted main trend information pair.

  For example, assume that the main expression extraction unit 11 extracts a1 and a2 as main unit expressions, b1 and b2 as main time expressions, and c1 and c2 as main item expressions. Depending on the combination of the extracted expressions, (a1, b1, c1), (a1, b1, c2), (a1, b2, c1), (a1, b2, c2), (a2, b1, c1) , (A2, b1, c2), (a2, b2, c1), and (a2, b2, c2), eight pairs of expressions are obtained. For example, the trend information pair extraction unit 12 extracts, as a trend information pair, an expression pair described in a place where an expression pair (a1, b1, c1) appears simultaneously from a target article group. The trend information pair extraction unit 12 extracts a trend information pair based on each expression pair by the same method. For example, a trend information pair extracted based on an expression pair having the largest number of extracted trend information pairs is set as a main trend information pair.

  In the description so far, each partial component is automatically performed. However, in the embodiment of the present invention, for example, the configuration of the main expression extraction unit 11 is removed, and the trend information pair extraction unit 12 is removed. However, a configuration may be adopted in which trend information pairs are extracted based on unit expressions, time expressions, and item expressions input by the user, and the main trend information pair extraction / display unit 13 displays the trend information pairs. Further, in the embodiment of the present invention, the main expression extraction unit 11 presents (displays) the extracted main unit expression, time expression, and item expression to the user, and the user is presented (displayed). Main unit representation, time representation, and item representation are selected from, and the trend information pair extraction unit 12 extracts a trend information pair based on the selected main unit representation, time representation, and item representation, The extraction / display unit 13 may display the trend information pairs as a graph. In the embodiment of the present invention, conversely, after the trend information pair is extracted to some extent through the processing by the main expression extraction unit 11 and the trend information pair extraction unit 12, the original newspaper article data and the manpower are extracted. In this case, the extracted information may be corrected and improved by comparing with the above and then displayed in a graph. The information extraction / display device 1 is constructed by being divided into a plurality of components, and a part of the configuration of the information extraction / display device 1 is replaced with human hands, or only a part of the information extraction / display device 1 is isolated. It is possible to use.

  FIG. 3 is a diagram illustrating a configuration example of the trend information pair extraction unit 12 in a case where the configuration of extracting trend information pairs using a machine learning technique is employed in the embodiment of the present invention. The trend information pair extraction unit 12 includes a teacher data storage unit 121, a solution-feature pair extraction unit 122, a machine learning unit 123, a learning result storage unit 124, an expression pair extraction unit 125, a feature extraction unit 126, a solution estimation unit 127, a trend. An information pair extraction unit 128 is provided.

  The teacher data storage unit 121 stores text data serving as teacher data used in the machine learning process. For example, the quantity expression is ai (i = 1, 2, 3,...), The time expression is bi (i = 1, 2, 3,...), And the item expression is ci (i = 1, 2, 3). ,...), Whether the ai, bi, ci pair (expression pair) appearing in the text data sentence is an expression pair to be extracted as a problem / trend information pair as teacher data Memorize cases that use information as a solution. Specifically, for every ai, bi, ci pair appearing in text data, either an expression pair to be extracted as a trend information pair (positive example) or an expression pair that should not be extracted (negative example) A tag indicating such a solution is manually attached. For example, (a1, b1, c1), (a1, b2, c1),... Are expression pairs configured based on the expressions a1, a2, b1, b2, c1, c2 in the text data shown in FIG. . . For each of (a2, b2, c2), a tag indicating a positive or negative solution is assigned.

That is, in the embodiment of the present invention, for example,
(A1, b1, c1)-solution "positive example"
(A1, b2, c1)-solution "negative example"
・
・
・
(A2, b2, c2)-solution "negative example"
A pair of expression pair and solution is generated.

  The solution-feature pair extraction unit 122 extracts a set of a solution and a set of features from an example of text data stored in the teacher data storage unit 121. The feature is information used in the machine learning process. The solution-feature pair extraction unit 122 uses, as a feature, for example, the distance (character or word) between ai and bi, bi and ci, and ai and ci for each expression pair to which a solution is given in certain text data. A range including an expression pair of ai, bi, and ci in text data, part-of-speech information before and after each of ai, bi, and ci. In addition, the answer-feature pair extraction unit 122 may include information such as whether ai, bi, and ci are included in the title of text data, and parts of speech that appear between ai and bi, bi and ci, and ai and ci. Information or information on whether ai includes a decimal point, bi is a year, month, day, or ci is a person name or place name may be used as a feature. In the embodiment of the present invention, the position information of ai, bi, and ci in an article may be a feature. For example, in articles such as newspapers, the first main expression that appears first is often important.

  The machine learning means 123 uses a supervised machine learning method to determine what kind of solution is likely to be generated from a set of the solution extracted by the solution-feature pair extraction means 122 and the feature set. learn. The learning result is stored in the learning result storage unit 124.

  The expression pair extraction unit 125 uses a main expression extracted by the main expression extraction unit 11 (for example, unit expression, time expression, item expression), ai (quantity expression) included in each article in the related article DB 14, All combinations (expression pairs) of three types of expressions, bi (time expression) and ci (item expression), are extracted. A combination of a numerical value appearing in an article connected to the unit expression and the unit expression is a quantity expression.

  The feature extraction unit 126 extracts a feature for each expression pair extracted by the expression pair extraction unit 125 by the same processing as the solution-feature pair extraction unit 122.

  The solution estimation unit 127 refers to the learning result of the learning result storage unit 124, and estimates the degree of the solution (classification destination) that is likely to be obtained in the case of a set of features for each expression pair.

  The trend information pair extraction unit 128 extracts, as a trend information pair, a presumption that the degree of expression pairs (positive examples) to be extracted as trend information pairs is high based on the estimation result of the solution estimation unit 127. .

  Here, a machine learning method by the machine learning means 123 will be described. The machine learning method prepares many sets of problem-solution pairs, learns them, learns what kind of solution the problem becomes, and uses the learning result to create a new problem. This is a method that makes it possible to guess the solution (for example, see the following references (5) to (7)).

Reference (5): Masaki Murata, Language Processing Based on Machine Learning, Faculty of Science and Engineering, Ryukoku University. Invited lecture. 2004.http: //www2.nict.go.jp/jt/a132/members/murata/ps/rk1-siryou.pdf
Reference (6): Japanese-English translation of tense aspect modality using support vector machine, Maki Murata, Mao, Kiyotaka Uchimoto, Hitoshi Isahara, IEICE Language Understanding and Communication Study Group NLC2000-78, 2001 Year.
Reference (7): CRL's efforts in the SENSEVAL2J dictionary task, Masaki Murata, Masao Uchiyama, Kiyotaka Uchimoto, Ma Aoi, Hitoshi Isahara, IEICE Language Understanding and Communication Study Group NLC2001-40, 2001.
In order to convey the problem situation to the machine, what kind of problem is required, features (elements constituting the problem with information used for analysis) are required. The problem is expressed by the feature. For example, in the problem of estimating the tense of Japanese sentence ending expressions, the problem: “He speaks.” --- If the solution “present” is given, an example of a feature is “He speaks.” . "" Speaking. "" Su "". "

  In other words, the machine learning method prepares many sets of feature set-solution pairs, performs learning, learns what kind of solution the feature set becomes, and uses the learning result. This is a method of extracting a set of features from a new problem and inferring a solution in the case of the feature.

  The machine learning unit 123 uses a technique such as a k-nearest neighbor method, a simple Bayes method, a decision list method, a maximum entropy method, or a support vector machine method as a machine learning method.

  The k-nearest neighbor method is a method for obtaining a classification destination (solution) by using the k most similar cases instead of the most similar case, and by majority decision of the k cases. k is a predetermined integer number, and generally an odd number between 1 and 9 is used.

  The Simple Bayes method is a method of estimating the probability of each classification based on Bayes' theorem and determining the classification having the highest probability value as a classification destination.

  In the simple Bayes method, the probability of outputting the classification a in the context b is given by the following equation (4).

Here, the context b is a set of features f _j (εF, 1 ≦ j ≦ k) set in advance. p (b) is the appearance probability of the context b. Here, since it is independent of the classification a and is a constant, it is not calculated. P (a) (where P is a tilde at the top of p) and P (f _i | a) are the probabilities estimated from the teacher data, respectively, and the appearance probability of class a, and the feature f for class a means the probability of having _i . When a value obtained by performing maximum likelihood estimation as P (f _i | a) is used, the value often becomes zero, and it may be difficult to determine the classification destination because the value of Equation (5) is zero. Therefore, smoothing is performed. Here, what smoothed using the following formula | equation (6) is used.

Here, freq (f _i , a) means the number of cases having the feature f _i and the classification a, and freq (a) means the number of cases having the classification a.

  The decision list method uses features and combinations of classification destinations as rules, stores them in the list in a predetermined priority order, and when input to be detected is given, from the highest priority in the list This is a method in which input data is compared with the feature of the rule, and the classification destination of the rule having the same feature is set as the classification destination of the input.

In the decision list method, the probability value of each classification is obtained using only one of the features f _j (εF, 1 ≦ j ≦ k) set in advance as a context. The probability of outputting classification a in a context b is given by

p (a | b) = p (a | fmax) Equation (7)
However, fmax is given by the following equation.

P (a _i | f _j ) (where P is a tilde at the top of p) is the rate of appearance of the classification a _i when the feature f _j is in the context.

In the maximum entropy method, when a set of preset features f _j (1 ≦ j ≦ k) is F, an expression (entropy) that satisfies a predetermined conditional expression (equation (9)) below ( In this method, the probability distribution p (a, b) when 10) is maximized is obtained, and the classification having the largest probability value is obtained from the probabilities of the respective classifications obtained according to the probability distribution.

However, A and B mean a set of classifications and contexts, and g _j (a, b) is a function that is 1 if the context b has a feature f _j and the classification is a, and is 0 otherwise. means. Further, P (a _i | f _j ) (where P is a tilde at the top of p) means the rate of appearance of (a, b) in the known data.

  Formula (9) is to obtain the expected value of the frequency of the output and feature pair by multiplying the probability p and the function g meaning the appearance of the pair of output and feature. And the expected value calculated based on the probability distribution calculated on the left side is the constraint, and entropy maximization (smoothing of the probability distribution) is performed to determine the probability distribution of the output and the context. Details of the maximum entropy method are described in the following references (8) and (9).

Reference (8): Eric Sven Ristad, Maximum Entropy Modeling for Natural Language, (ACL / EACL Tutorial Program, Madrid, 1997
(9): Eric Sven Ristad, Maximum Entropy Modeling Toolkit, Release 1.6beta, (http://www.mnemonic.com/software/memt,1998))
The support vector machine method is a method of classifying data composed of two classifications by dividing a space by a hyperplane.

  FIG. 5 shows the concept of margin maximization in the support vector machine method. In FIG. 5, a white circle means a positive example, a black circle means a negative example, a solid line means a hyperplane that divides the space, and a broken line means a surface that represents the boundary of the margin area. 5A is a conceptual diagram when the interval between the positive example and the negative example is narrow (small margin), and FIG. 5B is a conceptual diagram when the interval between the positive example and the negative example is wide (large margin). is there.

  At this time, if the two classifications consist of positive and negative examples, the larger the interval (margin) between the positive and negative examples in the learning data, the less likely it is to make an incorrect classification with open data. As shown in FIG. 5B, a hyperplane that maximizes this margin is obtained, and classification is performed using it.

  Basically, it is as described above. Usually, an extension of the method that the training data may contain a small number of cases in the inner area of the margin, or an extension that makes the linear part of the hyperplane nonlinear ( The one with the introduction of the kernel function is used.

  This extended method is equivalent to classification using the following discriminant function, and the two classes can be discriminated depending on whether the output value of the discriminant function is positive or negative.

Where x is the context (set of features) to be identified, and x _i and y _j (i = 1,..., L, y _j ∈ {1, -1}) mean the context and classification destination of the learning data. And the function sgn is
sgn (x) = 1 (x ≧ 0)
-1 (otherwise)
Each α _i is for maximizing the expression (12) under the constraints of the expressions (13) and (14).

  The function K is called a kernel function, and various functions are used. In this embodiment, the following polynomial is used.

K (x, y) = (x · y + 1) d Equation (15)
C and d are constants set experimentally. For example, C was fixed at 1 throughout all treatments. Moreover, two types of 1 and 2 are tried for d. Here, x _i where α _i > 0 is called a support vector, and the portion taking the sum of Expression (11) is usually calculated using only this case. That is, only actual cases called support vectors are used for actual analysis.

  The details of the extended support vector machine method are described in the following references (10) and (11).

Reference (10): Nello Cristianini and John Shawe-Taylor, An Introduction to Support Vector Machines and other kernel-based learning methods, (Cambridge University Press, 2000)
Reference (11): Taku Kudoh, Tinysvm: Support Vector machines, (http://cl.aist-nara.ac.jp/taku-ku//software/Tiny SVM / index.html, 2000)
The support vector machine method handles data with two classifications. Therefore, when handling cases with three or more classifications, a pair-wise method or a one-VS rest method is usually used in combination with this.

  In the pairwise method, in the case of data having n classifications, every pair (n (n-1) / 2) of two different classification destinations is generated, and a binary classifier indicates which is better for each pair. That is, it is obtained by the support vector machine method processing module and finally obtains the classification destination by majority decision of the classification destination by n (n−1) / 2 binary classification.

  For example, when there are three classification destinations a, b, and c, the one VS rest method generates three sets of classification destination a and other, classification destination b and other, classification destination c and other, The learning process is performed on the set of the support vector machine method. Then, in the estimation process based on the learning result, the learning results of the three sets of support vector machines are used. See how the three support vector machines are estimated as candidates to be estimated. Of the three support vector machines, it is the non-other classification target and the separation plane of the support vector machine. This is a method for obtaining a classification destination of a thing farthest from the object. For example, when a candidate is farthest from the separation plane in the support vector machine created by the learning process of “classification destination a and other”, the candidate classification destination is estimated as a.

  The method of determining the level of the solution (classification destination) that is likely to be the solution (classification destination) for each expression pair estimated by the solution estimation unit 127 differs depending on various methods used by the machine learning unit 123 as a machine learning method.

  For example, in the embodiment of the present invention, when the machine learning means 123 uses the k-nearest neighbor method as a machine learning technique, the machine learning means 123 sets the feature data extracted from the cases among the cases of the teacher data. The similarity between cases based on the ratio of overlapping features (the number of the same features) is defined, and the defined similarity and the case are stored in the learning result storage means 124 as learning result information. Keep it.

  Then, when a new expression pair (candidate) is extracted by the expression pair extraction unit 125, the solution estimation unit 127 refers to the similarity and the case defined in the learning result storage unit 124, and the expression pair extraction unit 127 For the candidates of the expression pairs extracted by 125, k cases are selected from the cases in the learning result storage means 124 in descending order of the similarity of the candidates, and the classification destinations determined by the majority vote in the selected k cases are selected. Estimated as the classification target (solution) of the expression pair candidate. That is, in the solution estimation means 127, the degree of what kind of solution (classification destination) is likely to be obtained for each expression pair is the number of votes of the majority vote in the selected k cases, here “to be extracted”. The number of votes obtained by classification.

  When the simple Bayes method is used as the machine learning method, the machine learning unit 123 stores, in the learning result storage unit 124, a set of a solution of the case and a set of features as learning result information for the example of the teacher data. Remember. Then, the solution estimation means 127, when a new expression pair (candidate) is extracted by the expression pair extraction means 125, based on the combination of the learning result information solution and the feature set in the learning result storage means 124. Based on Bayes' theorem, the probability of becoming each classification in the case of the feature pair of the expression pair candidates acquired by the feature extraction means 126 is calculated, and the classification having the largest probability value is selected as the candidate of the expression pair. It is estimated as the classification (solution) of the features of. That is, in the solution estimation means 127, the degree of the likelihood of becoming a certain solution in the case of a set of features of expression pair candidates is set as the probability of becoming each classification, here, the probability of becoming the classification “to be extracted”.

  When the decision list method is used as the machine learning method, the machine learning unit 123 stores, in the learning result storage unit 124, a list in which rules of features and classification destinations are arranged in a predetermined priority order with respect to examples of teacher data. Remember. Then, when a new expression pair (candidate) is extracted by the expression pair extraction unit 125, the solution estimation unit 127 selects candidate expression pairs extracted in descending order of priority in the list of the learning result storage unit 124. The feature is compared with the feature of the rule, and the classification destination of the rule having the identical feature is estimated as the candidate classification destination (solution). That is, the solution estimation means 127 assigns the degree of the likelihood of becoming a solution in the case of a set of candidate features of the expression pair to a predetermined priority or a numerical value or scale corresponding thereto, in this case, “to be extracted”. Priority in the list of probabilities.

  When the maximum entropy method is used as the machine learning method, the machine learning means 123 specifies a class that can be a solution from the example of the teacher data, and maximizes an expression that satisfies a predetermined conditional expression and shows entropy. A probability distribution consisting of a set of features and a class that can be a solution is obtained and stored in the learning result storage means 124. When a new expression pair (candidate) is extracted by the expression pair extraction unit 125, the solution estimation unit 127 uses the probability distribution of the learning result storage unit 124 to identify the features of the extracted expression pair candidate. The probability of the classification that can be the solution for the set of is determined, the classification that can be the solution having the largest probability value is identified, and the identified classification is estimated as the candidate solution. That is, in the solution estimation means 127, the degree of the likelihood of becoming a certain solution in the case of a set of features of expression pair candidates is set as the probability of becoming each classification, here, the probability of becoming the classification “to be extracted”.

  When the support vector machine method is used as the machine learning method, the machine learning unit 123 specifies a class that can be a solution from the example of the teacher data, divides the class into a positive example and a negative example, A hyperplane that maximizes the interval between the positive and negative examples of a case and divides the positive and negative examples by a hyperplane in a space whose dimension is a set of case features according to a predetermined execution function using Is stored in the learning result storage means 124. Then, when a new expression pair (candidate) is extracted by the expression pair extraction unit 125, the solution estimation unit 127 uses the hyperplane of the learning result storage unit 124 to identify the feature of the extracted expression pair candidate. Whether the set is on the positive example side or the negative example side in the space divided by the hyperplane is specified, and the classification determined based on the specified result is estimated as the candidate solution. That is, in the solution estimation means 127, the degree of the likelihood of being a solution in the case of a set of candidate expression pairs is the distance from the separation plane to the space of the positive example (expression pair to be extracted). . More specifically, when the expression pair to be extracted is a positive example and the expression pair that should not be extracted is a negative example, the case located in the space on the positive example side with respect to the separation plane is referred to as “example to be extracted”. The distance from the separation plane of the case is determined as the degree of the case.

FIG. 6 is a diagram showing an example of an information extraction / display processing flow in the embodiment of the present invention. First, the information extraction / display apparatus 1 extracts a main expression from an article group in the related article DB 14 (step S1). Next, the information extraction / display apparatus 1 extracts a trend information pair using the extracted main expression (step S2). Then, the information extraction / display device 1 displays the extracted information trend pair (step S3).
(Experiment and discussion)
(1) Main Expression Extraction Using the information extraction / display apparatus 1 of the present invention, an experiment of main expression extraction was performed. Okapi's TF term formula (formula (1)) is used, and item representation uses a method in which TF _i is the product of the number of occurrences of the expression and the character string length of the expression. In this experiment, the main expressions were extracted from each of the articles related to the typhoon, the articles related to the major leagues, and the articles related to the political trend. The extraction result is shown in FIG. In the table shown in FIG. 7, the top five expressions having large values of the expression of the Okapi TF term are shown.

  As shown in FIG. 7, the main expressions in each field are extracted well. For example, in the case of typhoon data, the main item expression "typhoon" and the unit expression "number" indicating the number of typhoon can be extracted. In the case of major league data, we have collected data when Maguire and Sosa's home run struggle became popular, and we have successfully extracted Maguire and Sosa at the top. In addition, “No.”, “Book”, etc., which are the main unit expressions in home run battles, can be extracted well. In the case of political trend data, “Category support rate” can be successfully extracted as an item expression and “%” as a unit expression.

By the way, in the typhoon data, the expression “photo description” appears at the top. From this, it is probable that the typhoon data was accompanied by a picture of the damage caused by the typhoon.
(2) Graphing trend information Using the information extraction / display device 1 of the present invention, an experiment for graphing trend information was performed. This experiment uses the unit expression, time expression, and item expression with the highest score value calculated by the main expression extraction unit 11, through the processing of the trend information pair extraction unit 12 and the main trend information pair extraction / display unit 13. Graphed. The created graphs are shown in FIGS. 8 (A) to 8 (C). In this experiment, Excel was used for graphing. Here, the time axis is displayed for each obtained time expression, but may be displayed at an actual time interval. As time expression, when only the information up to the day and month can be obtained automatically, the month and year information was manually added.

  In the typhoon data, the main expression extraction unit 11 extracts “No.” as a unit expression, “day” as a time expression, and “typhoon” as an item expression. A graph was created using this. The trend information pair extraction unit 12 extracts a place where these three expressions appear simultaneously. In the extracted data, a graph was created with the horizontal axis representing time and the vertical axis representing the number in front of the unit number “No.”. The typhoon data is only in September and October in the related articles in the related article DB 14, and information on other periods is unknown, but the graph of FIG. 8A for typhoon data is in September. It will help you know when and how many typhoons came in October. A comparison of the data for 1998 and 1999 shows that the number of typhoons was higher in 1999.

  In the major league data, the main expression extraction unit 11 took out “No.” as the unit expression, “Day” as the time expression, and “Maguire” as the item expression. A graph was created using this. The trend information pair extraction unit 12 extracts a place where these three expressions appear simultaneously. In the extracted data, a graph was created with the horizontal axis representing time and the vertical axis representing the number in front of the unit number “No.”. Major league data is the related article in the related article DB 14 that originally has only data from August onwards, and information on other periods is not known, but the graph in FIG. You can see how he was hitting the home run.

  In the data on political trends, the main expression extraction unit 11 extracted “%” as a unit expression, “month” as a time expression, and “Cabinet support rate” as an item expression at the highest level. A graph was created using this. The trend information pair extraction unit 12 extracts a place where these three expressions appear simultaneously. In the extracted data, a graph was created with the horizontal axis representing time and the vertical axis representing the number in front of the unit expression “%”. For example, related articles related to the cabinet support rate in the related article DB 14 are data for 1998 and 1999, and the graph of FIG. 8C is a graph showing the cabinet support rate of the Kominato Cabinet. Although the support rate is undulating, it can be seen that it is generally on an upward trend.

Although the experimental results are omitted here, the main expression extraction unit 11 extracts a plurality of unit expressions, time expressions, and item expressions, and the trend information pair extraction unit 12 in the data for all combinations of the plurality of expressions. We are also experimenting with a framework that extracts multiple types of trend information pairs using, and judges and extracts as more useful trend information pairs of trend information pairs that can be extracted in the multiple types of trend information pairs. This framework was useful when useful unit expressions, time expressions, and item expressions could not be extracted at the top level.
(3) Sentence Extraction and Highlight Display Using the information extraction / display apparatus 1 of the present invention, experiments on sentence extraction and highlight display regarding trend information were performed. Here, we experimented with related articles related to typhoons. This experiment uses the unit expression, time expression, and item expression extracted at the highest level by the main expression extraction unit 11, through processing by the trend information pair extraction unit 12 and the main trend information pair extraction / display unit 13, Extraction and highlighting were performed. That is, the sentence including the trend information pair extracted by the trend information pair extraction unit 12 is extracted from the related article group in the related article DB 14 by the main trend information pair extraction / display unit 13, and the trend information is emphasized in the extracted sentence. displayed.

  For example, the unit expression, time expression, and item expression are “No.”, “Day”, and “Typhoon”. In sentence extraction, sentences in which these three expressions appear simultaneously are extracted. Then, the three expressions are highlighted in the sentence. When there are a plurality of expressions in the same sentence, for example, the first appearing is highlighted with a double line and the others are highlighted with a single line. The result is shown in FIG. In the embodiment of the present invention, a configuration may be adopted in which the above three expressions are displayed with appropriate color coding. In the example of highlighting shown in FIG. 9, the extracted time expression and numerical expression are added in front of the extracted sentence.

  The extracted sentence clearly shows the state of the typhoon at that time, and it seems that the sentence having the same effect as the important sentence extraction in the summary research can be extracted. In other words, the name of the place through which the typhoon passed and the damage situation are also described, and an important description about the typhoon is included in the extracted sentence. It turns out that an important sentence can be extracted only by taking out a sentence in which three data of unit expression, time expression, and item expression appear simultaneously.

  In addition, the seventh data in the figure includes a plurality of data of typhoon No. 7 and typhoon No. 8. In addition to the extracted information, if there is a main expression currently focused on, it is also a single underline. By highlighting with, you can immediately see that the data is in the data. In addition, the system may mistake the combination of unit expression, time expression, and item expression to be extracted. This highlighting is useful for finding the error early. Here, only the extracted sentence is highlighted, but there is a possibility that the sentence to be extracted remains in the article. If the same highlighting is applied to the entire article, such a leak may be extracted. Thus, a configuration may be adopted in which the entire original article is highlighted.

  The present invention can also be implemented as a program that is read and executed by a computer. The program for realizing the present invention can be stored in an appropriate recording medium such as a portable medium memory, a semiconductor memory, or a hard disk, which can be read by a computer, provided by being recorded on these recording media, or communication. It is provided by transmission / reception using a network via an interface.

It is a figure which shows an example of the system configuration | structure of this invention. It is a figure which shows the structural example of a main unit expression extraction part, a main time expression extraction part, and a main item expression extraction part. It is a figure which shows the structural example of a trend information pair extraction part. It is a figure which shows an example of text data. It is a figure which shows the concept of margin maximization of a support vector machine method. It is a figure which shows an example of an information extraction and a display processing flow. It is a figure which shows the extraction result of main expression. It is a figure which shows an example of graphing of trend information. It is a figure which shows the trend information highlighted.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Information extraction / display apparatus 11 Main expression extraction part 12 Trend information pair extraction part 13 Main trend information pair extraction / display part 14 Related article DB
111 Main Unit Expression Extraction Unit 112 Main Time Expression Extraction Unit 113 Main Item Expression Extraction Unit 121 Teacher Data Storage Unit 122 Solution-Feature Pair Extraction Unit 123 Machine Learning Unit 124 Learning Result Storage Unit 125 Expression Pair Extraction Unit 126 Feature Extraction Unit 127 Solution Estimating means 128 Trend information pair extracting means 200 Unit expression extracting means 201, 301, 401 Score value calculating means 202 Main unit expression extracting means 300 Time expression extracting means 302 Main time expression extracting means 400 Item expression extracting means 402 Main item expression extracting means

Claims (7)

An information extraction / display device that extracts and displays trend information,
Unit expression extracting means for extracting a unit expression that appears frequently in the entire article group as a main unit expression from an article group related to a certain field , and a time expression that appears frequently in the entire article group from the article group The main unit expression, the main time expression, the time expression extracting means for extracting as the main time expression, and the item expression extracting means for extracting, as the main item expression, the item expression that appears frequently in the entire article group from the article group, Main expression extraction means for extracting main item expressions as main expressions ;
Based on the main expression extracted by the main expression extracting means, the main unit when the main unit expression, the main time expression, and the main item expression appear from the article group at the same time without any breaks in the document. Trend information pair extraction means for extracting a combination of expression, main time expression and main item expression as a trend information pair;
An information extraction / display apparatus comprising: display means for displaying trend information pairs extracted by the trend information pair extraction means . The information extraction / display apparatus according to claim 1 ,
The trend information pair extraction means extracts features from teacher data storing what should be extracted as the trend information pair given in advance and what should not be extracted, and tends to become a trend information pair at any feature Machine learning means for storing the learning result data learned in the learning result data storage means,
The feature is extracted from the trend information pair obtained in claim 1 and the learning result data is used to estimate whether the trend information pair is likely to become a trend information pair. An information extraction / display apparatus characterized by extracting only large trend information pairs as trend information pairs . In the information extraction / display device according to claim 1 or 2 ,
The main expression extracting means extracts a plurality of main expressions,
The display means extracts a main trend information pair from a plurality of types of trend information pairs extracted by the trend information extraction means based on the extracted main expression, and displays the extracted main trend information pairs An information extraction / display device characterized by this. In the information extraction and display device according to any one of claims 1 to 3,
The trend information pair extraction means further extracts a trend information pair from the article group based on a selected main expression among the main expressions extracted by the main expression extraction means. -Display device. In the information extraction and display device according to any one of claims 1 to 4,
A keyword input means for inputting keywords;
An article group extracting unit that extracts an article group related to the input keyword from bibliographic data stored in a storage unit;
The information extraction / display apparatus, wherein the main expression extraction unit extracts the main expression from the article group extracted by the article group extraction unit. In the information extraction and display device according to any one of claims 1 to 5,
The display means displays a graph using the trend information pair extracted by the trend information pair extraction means as a graph using the main time expression on the horizontal axis and the numerical value of the main unit expression on the vertical axis. -Display device. In the information extraction and display device according to any one of claims 1 to 6,
The display unit extracts a sentence including the trend information pair extracted by the trend information pair extraction unit from the article group, and highlights the trend information pair in the extracted sentence. Extraction and display device.

Images