JP2005063298A - Document processing unit and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To properly extract a label and keywords of a document group. <P>SOLUTION: A phrase extracting part 1 extracts an important phrase from each sentence by morphological analysis. A phrase importance score calculating part 2 calculates a phrase importance score per each phrase. An inclusive relation analyzing part 3 creates a table indicating inclusive relations of the important phrases. A label extraction score calculating part 4 newly calculates a label extraction score from the phrase importance score of each phrase so that a label extraction score of an included phrase is higher than a label extraction score of an inclusive phrase. A keyword extraction score calculating part 5 calculates a keyword extraction score by adjusting the phrase importance score so that the included phrase is extracted as a keyword. A label selecting part 6 selects a phrase with the highest label extraction score as a label. A keyword selecting part 7 selects several phrases with high phrase importance scores from the top as keywords. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to document processing for extracting a label representing the contents of a document set in one word and a keyword representing an outline of the document set.

  In recent years, digitization of documents has progressed, and a large number of documents have been digitized and released or shared. It is very difficult to find a document that the user needs from such a large amount of electronic documents. For this reason, various methods for searching for documents required by the user from a large number of documents have been considered.

  One of them is a method of processing a document group by classifying the document according to the content of the document and grouping it as a document group. In order to classify a document into a document group, there are a method of automatically classifying a document in addition to manual classification, and a case where a document group is obtained as a search result.

  In this way, in order to distinguish them from other document groups, important words and phrases in the document group are extracted and displayed as keywords, and the contents of the document group are summarized. Displayed, and names and labels are assigned to documents.

  However, it is difficult to express the contents of a group of documents in a single word only by displaying a keyword, and it is difficult to grasp what the entire contents are specifically by using only a label.

  As means for solving this problem, a label indicating the name of the document group and a plurality of keywords indicating the contents of the document group are assigned. In this case, the frequency of words extracted from the documents in the document group is generally calculated, the keyword candidate phrases are output, the word with the highest frequency of occurrence is used as a label, and the top words of the remaining words are counted. The method that uses or as a keyword is taken.

  For example, in ConceptBase Clustering (trademark) of Just System Co., Ltd., the top one of the keywords is selected as the label.

  However, labels and keywords have different roles. The label represents the entire document group, and is suitable for a concept common to important concepts appearing in the document group, and the first one of the keywords is not necessarily suitable. On the other hand, the keyword has a role of explaining the contents of the document group to the user in an easy-to-understand manner, and a more specific phrase is suitable.

  Therefore, it is necessary to change the method of extracting labels and keywords.

Although Patent Document 1 describes that words are clustered and each cluster is displayed together with the most important main keyword, this does not extract a document group label.
JP 2001-325272 A

This invention was made in consideration of the above circumstances,
It is intended to provide.

  The present invention has been made in order to solve the above-described problems of the prior art, and for a document group, a label representing the contents of a document in one word and a keyword for explaining the contents of the document in more detail are provided. It is an object of the present invention to provide a document processing apparatus that enables extraction using different methods suitable for each.

  According to this invention, in order to achieve the above-mentioned object, the configuration as described in the claims is adopted. Here, prior to describing the invention in detail, supplementary explanations of the claims will be given.

  That is, according to one aspect of the present invention, in order to achieve the above-described object, a document processing apparatus that extracts a label that represents the contents of a document set with one word and one or more keywords that represent an outline of the document set. N: a phrase extracting unit that extracts a candidate for a label and a keyword from each document in the document set; a phrase importance score calculating unit that calculates a phrase importance score representing the importance of the phrase for each phrase; An inclusion relation analyzing means for analyzing a morphological inclusion relation between the phrases; and a phrase included in another phrase based on the inclusion relation analyzed by the inclusion relation analyzing means for the phrase importance score A label extraction score calculating means for calculating a label extraction score by adjusting so as to add points; an inclusion analyzed by the inclusion relation analyzing means for the word importance score; A keyword extraction score calculating means for calculating a keyword extraction score by adjusting so that a deduction is performed on a phrase including another phrase based on the relationship; one label is selected from the phrases according to the label extraction score; Label selecting means for selecting; and keyword selecting means for selecting a keyword based on the keyword extraction score are provided.

  In this configuration, the likelihood that a word to be included is selected as a label based on the inclusion relationship on the form of the word is increased, and the likelihood that the included word is selected as a keyword is increased. Phrases that are concepts common to important concepts can be easily selected as labels, and words with specific meanings can be easily selected as keywords.

  The inclusion relation analyzing means analyzes, for example, the inclusion relation of character strings and the inclusion relation of word strings.

  Further, the keyword selection means may not select included words as keywords.

  Further, a portion corresponding to the label in the keyword may be displayed separately from other portions.

  Further, means for classifying a document group may be added to the document processing apparatus.

  According to another aspect of the present invention, there is provided a document processing apparatus for extracting a label that represents the contents of a document set with one word: a phrase extracting unit that extracts a phrase that is a candidate for a label from each document in the document set; A phrase importance score calculating means for calculating a phrase importance score representing the importance of the phrase for each word; inclusion relation analyzing means for analyzing an inclusion relation between the phrases; and the phrase importance score; Label extraction score calculation means for adjusting and calculating a label extraction score based on the inclusion relation analyzed by the inclusion relation analysis means; and label selection means for selecting a label based on the label extraction score are provided. ing.

  Even in this configuration, it is possible to appropriately select a word / phrase as a label based on the inclusion relation of the word / phrase.

  The inclusion relationship between the words is an inclusion relationship in terms of a phrase or a semantic inclusion relationship. The inclusion relation in terms of the phrase is, for example, an inclusion relation of a character string or an inclusion relation of a word string. The semantic inclusion relationship can be analyzed using a dictionary, for example.

  The present invention can be realized not only as an apparatus or a system but also as a method. Of course, a part of the invention can be configured as software. Of course, software products used to cause a computer to execute such software are also included in the technical scope of the present invention.

  These and other aspects of the invention are set forth in the appended claims and will be described in detail below with reference to examples.

  According to the present invention, there is an effect that appropriate labels and keywords can be extracted from a document group including one or a plurality of documents. By assigning such labels and keywords, it is possible to correctly identify a document group.

  Examples of the present invention will be described below.

  FIG. 1 is a block diagram showing the entire document processing apparatus according to the first embodiment of the present invention. In FIG. 1, a document processing apparatus includes a phrase extraction unit 1, a phrase importance score calculation unit 2, an inclusion relation analysis unit 3, a label extraction score calculation unit 4, a keyword extraction score calculation unit 5, a label selection unit 6, and a keyword selection unit. 7 and the display output part 8 etc. are comprised. Specifically, a computer program for realizing the functions of these units is executed by a predetermined computer or a group of computers. Of course, some or all of these units may be configured by hardware.

  In this embodiment, it is assumed that a document set including documents including at least text is configured in advance. These document groups are obtained by automatically classified results or retrieved results.

  The phrase extraction unit 1 performs text morphological analysis on each document in the document group, and extracts an important phrase from each sentence. These words / phrases are not extracted by morphological analysis, but may be extracted using other methods.

  The phrase importance score calculation unit 2 calculates a phrase importance score indicating the importance of the phrase for each phrase with respect to the phrase obtained by the phrase extraction unit 1. As a method for calculating the phrase importance score, a conventionally used tfidf (Term Frequency / Inverse Document Frequency) method can be used, but the method is not limited to this.

For example, the weight of an element word included in a certain document group can be expressed by the following formula applying the mutual information amount of the document group with respect to the entire target document, and can also be used as a phrase importance score.

  The inclusion relationship analysis unit 3 analyzes the inclusion relationship of words and creates a table indicating the inclusion relationship of important words.

  The label extraction score calculation unit 4 receives the inclusion relationship of the three words in the inclusion relationship analysis unit, and among the words in the inclusion relationship, the label extraction score of the included word is higher than the label extraction score of the included word Thus, a label extraction score is newly calculated from the phrase importance score of each phrase.

  The keyword extraction score calculation unit 5 adjusts the phrase importance score so as to receive the inclusion relation of the words in the inclusion relation analysis part 3 and extract the inclusion words as keywords from the words in the inclusion relation. An extraction score is calculated.

  The label selection unit 6 selects the label extraction score calculated by the label extraction score calculation unit 4 with the highest score as a label.

  The keyword selection unit 7 selects, as keywords, some of the higher-ranked scores from the phrase importance scores calculated by the phrase importance score calculation unit 5.

  The display output unit 8 displays and outputs the label selected by the label selection unit 6 and the keyword selected by the keyword selection unit 7.

  When each document in the document group is input, the phrase extraction unit 1 first extracts a phrase from the text in the document and extracts an important phrase. As a word to be extracted, only an independent word or only a noun may be extracted. There are methods such as extracting these words by morphological analysis, but other methods may be used. These extracted phrases are designated as important phrases.

  Next, the phrase importance score calculation unit 2 calculates the phrase importance score of each extracted phrase. An example of an important phrase and phrase importance score is shown in FIG. In FIG. 3, “Husein”, “Weapons of Mass Destruction”, “Inspection”, “Nuclear Weapons”, “Hussein Administration”, “United Nations”, “Weapons” are extracted as words, and the word importance score is “0.4,0” for each word. .3, 0.2, 0.2, 0.1, 0.1, 0.1 ".

  Next, the inclusion relationship analysis unit 3 analyzes the inclusion relationship of the important words. At this time, a table for holding the word relationship in the inclusion relationship is created. FIG. 4 shows a table created by the inclusion relationship analysis unit for the example of FIG.

  In the label extraction score calculation unit 4, the label extraction score of the included word / phrase is higher than the label extraction score of the included word / phrase among the words / phrases in the inclusion relationship based on the table created by the inclusion relationship analysis unit 3. As described above, the phrase importance score of each word is adjusted to calculate the label extraction score. Here, the label extraction score is calculated by adding the phrase importance score of the included phrase and the phrase importance score of the included phrase.

  FIG. 2-A shows a flowchart for calculating the label extraction score.

  The label extraction score calculation unit 4 calculates the label extraction score of the phrase based on the table created by the inclusion relationship analysis unit 3. In the case of a phrase having an inclusion relationship, the phrase importance score of the included phrase and the phrase importance score of the included phrase are added to obtain a label extraction score for the included phrase (S13, S14). Each step S10 to S16 in FIG. 2A is clear from the figure and will not be described in detail.

  When the label extraction score is calculated for the example of FIG. 3, Hussein and the Hussein administration are in an inclusive relationship. Therefore, the Hussein administration's phrase importance score of 0.1, which is the included phrase, and Hussein's phrase importance of the included phrase Add to degree score 0.4. As a result, the label extraction score for Hussein is 0.5.

  Similarly, since weapons, weapons of mass destruction, and nuclear weapons are also in an inclusive relationship, the phrase importance score of weapons of the included words and the phrase importance score of weapons of mass destruction and nuclear weapons are added. As a result, the weapon label extraction score is 0.6.

  When the word is not included, such as “inspection” or “UN”, the phrase importance score is given to the label extraction score. The result of calculating the label extraction score in this manner is shown in FIG.

  The keyword extraction score calculation unit 5 calculates a keyword extraction score for the important phrase extracted by the phrase analysis unit.

  FIG. 2-B shows a flowchart for calculating the keyword extraction score.

  The keyword extraction score calculation means calculates the keyword extraction score of the phrase based on the table created by the inclusion relationship analysis unit 3. In the case of words in an inclusive relationship, a keyword extraction score for a word that includes the higher score of the included words and the word importance score of the included words is set (S23, S24). Each step S20 to S26 of FIG. 2B is also clear from the figure, and will not be described in detail.

  Since it is appropriate that the word / phrase as a keyword indicates the contents of the document group as specifically as possible, the keyword extraction score calculation unit 5 includes the word / phrase including the higher score of the word / phrase included and the score of the word / phrase included. And a keyword extraction score is calculated.

  FIG. 6 shows the keyword extraction score calculated for the example of FIG. In FIG. 6, “Husein”, “Weapons of mass destruction”, “Inspection”, “Nuclear weapons”, “Husein administration”, “United Nations”, “Weapons” are listed as important words. , 0.3, 0.2, 0.2, 0.1, 0.1, 0.1 ".

  Here, the “Hussein administration” includes “Hussein”. In this case, in order to select a more specific word as a keyword, the keyword extraction score of “Hussein administration” is assigned the higher one (0.4) of the phrase importance score of “Hussein” and “Hussein administration”. “Weapons of mass destruction” and “weapons”, “nuclear weapons” and “weapons” are also in an inclusive relationship. “Mass destruction weapons” is a phrase that encompasses “weapons”, and gives the keyword extraction score of “mass destruction weapons” the highest importance score (0.3) of “mass destruction weapons” and “weapons” . The same applies to “nuclear weapons” and “weapons”, and the keyword extraction score for “nuclear weapons” is given the one with the highest phrase importance score of “nuclear weapons” and “weapons” (0.2).

  The label selection unit 6 selects “weapon” having the highest label extraction score as a label.

  Since the keyword selection unit 7 does not select the label “weapon” as a keyword, “hussein”, “mass destruction weapons”, “inspection”, “nuclear weapons”, “hussein administration”, and “UN” are extracted as keywords.

  In keyword extraction, all may be keywords, or a predetermined number or a score higher than a predetermined score may be extracted as keywords. Moreover, the keyword selection part 7 may select only the phrase to include, when there exists a phrase with an inclusion relationship. In this example, “Hussein” is not selected.

  FIG. 7 shows changes in the phrase importance score, the label extraction score, and the keyword extraction score.

  The labels and keywords extracted in this way are displayed and output by the display output unit 8. An example of the output method is shown in FIGS. 8-A, 8-B, and 8-C.

  The method for assigning labels and keywords to one document group has been described above as the first embodiment of the present invention.

  Next, as a second embodiment of the present invention, a case will be described in which a group of documents is clustered and labels and keywords are assigned.

  FIG. 9 is a block diagram showing the entire document processing apparatus according to the second embodiment. In FIG. 9, the document processing apparatus includes a phrase extraction unit 1, a phrase importance score calculation unit 2, an inclusion relation analysis unit 3, a label extraction score calculation unit 4, a keyword extraction score calculation unit 5, a label selection unit 6, and a keyword selection unit. 7, a display output unit 8, a label / keyword holding unit 9, and the like.

  In this embodiment, when the document group is clustered, the label / keyword holding unit 9 secures an area for holding a label and a keyword set for each cluster.

  When each document of the first cluster is input, the phrase extraction unit 1 first extracts the phrase of the text in the document and extracts the important phrase.

Next, the phrase importance score calculation unit 2 calculates the phrase importance score of each extracted phrase. An example of an important phrase and phrase importance score is shown in FIG. In FIG.
"Energy saving""Powerconsumption""Printer""Consumption""High image quality""Environment""Energy"
Are extracted as words, and the word importance score is 0.4, 0.4, 0.3, 0.2, 0.1, 0.1, 0.1 for each word.
And so on.

  The inclusion relationship analysis unit 3 analyzes the inclusion relationship of words and creates a table indicating the inclusion relationship of important words.

  In the label extraction score calculation unit 4, from the inclusion relationship of the words analyzed by the inclusion relationship analysis unit 3, among the words in the inclusion relationship, the label extraction score of the included word is higher than the label extraction score of the word included. Adjust so that Here, the label extraction score of the included phrase is calculated by adding the phrase importance score of the included phrase and the importance score of the included phrase. FIG. 11 shows a table representing the inclusion relations of important words and phrases created by the inclusion relation analysis unit 3 for the example of FIG.

  When the label extraction score is calculated for the example of FIG. 10, since “energy saving” and “energy” are in an inclusive relationship, the phrase importance score of “energy saving”, which is an included word, is included as 0.4. Add a phrase importance score of 0.1 for the phrase “energy”. As a result, the label extraction score for “energy” is 0.5.

  Similarly, since “power consumption” and “consumption” are also in an inclusive relationship, the phrase importance score 0.4 of “power consumption” that is an included phrase and the phrase importance score 0 of “consumption” of the included phrase are 0. .2 is added. As a result, the label extraction score for “consumption” is 0.6.

  The result of calculating the label extraction score in this way is shown in FIG.

  Since the label selection unit 6 selects the label with the highest label extraction score as the label, in FIG. 12, “consumption” with the highest label extraction score is selected as the label.

  Next, keyword extraction is performed from the important phrases extracted by the phrase extraction unit 1. It is appropriate that the word / phrase to be a keyword is as specific as possible showing the contents of the document group. Therefore, if there is a word / phrase in an inclusive relationship, the word / phrase to be included is selected regardless of the word / phrase importance score. Therefore, the keyword extraction score calculation unit 5 sets a keyword extraction score for a phrase including the higher score among the included phrases and the score of the included phrases.

  FIG. 13 shows the keyword extraction score calculated for the example of FIG. Here, since “energy saving” is a phrase that includes “energy”, the higher one (0.4) of the phrase importance score of “energy saving” and “energy” is assigned to the keyword extraction score of “energy saving”. . Since “power consumption” is a phrase including “consumption”, the higher word importance score (0.4) of “power consumption” and “consumption” is assigned to the keyword extraction score of “power consumption”.

The keyword selection unit 7 does not select a word selected as a label as a keyword. Also, if there is an inclusion relationship here, if you select only the words to include,
"Energy saving""Powerconsumption""Printer""High image quality""Environment"
Is extracted as a keyword.

  FIG. 14 shows changes in the phrase importance score, the label extraction score, and the keyword extraction score.

  The extracted label and keyword are stored by the label / keyword holding unit 9.

  Next, when each document of the second cluster is input, the phrase extracting unit 1 first extracts the phrase of the text in the document and extracts the phrase that is a keyword candidate.

  Next, the phrase importance score calculation unit 2 calculates the phrase importance score of each extracted phrase.

  An example of an important phrase and phrase importance score is shown in FIG. In FIG. 15, “business process”, “activity”, “document information”, “change”, “theme”, “intellectual activity”, and “viewpoint” are extracted as words, and the word importance score is 0.3,0 for each word. .3, 0.2, 0.2, 0.2, 0.1, 0.1 and so on.

  FIG. 16 shows a table representing the inclusion relationship of the key words created by the inclusion relationship analysis unit 3 for the example of FIG. When the label extraction score is calculated for the example of FIG. 15, since “activity” and “intellectual activity” are in an inclusive relationship, the phrase importance score 0.1 of “intellectual activity” that is an inclusive phrase is The phrase importance score 0.3 of “activity” of the included phrase is added. As a result, the label extraction score for “activity” is 0.4.

  The result of calculating the label extraction score in this way is shown in FIG.

  Since the label selection unit 6 selects the label with the highest label extraction score as the label, in FIG. 17, the “activity” with the highest label extraction score is selected as the label.

  FIG. 18 shows the keyword extraction score calculated for the example of FIG. Here, “intellectual activity” is a phrase that includes “activity”. In this case, the higher one (0.3) of the phrase importance score of “intellectual activity” and “activity” is assigned to “intellectual activity”.

  Since the keyword selection unit 7 does not use the word or phrase as the label as a keyword, it extracts “business process”, “document information”, “change”, “theme”, “intellectual activity”, and “viewpoint” as keywords.

  FIG. 19 shows changes in the phrase importance score, the label extraction score, and the keyword extraction score.

  The extracted label and keyword are stored by the label / keyword holding unit 9.

  Thereafter, the same processing is performed for other clusters.

  The labels and keywords extracted in this way and held by the label / keyword holding unit 9 are displayed and output by the display output unit 8. An example of the output method is shown in FIGS. 20-A, 20-B, and 20-C.

  In this case, there is a possibility that the same label is given to a plurality of document groups, and it is considered that the contents of the document groups are difficult to distinguish. Therefore, when clustering, adjustment is made so that the same label is not given. Refer to Japanese Patent Application No. 2002-076919 for this method.

  In this way, by extracting one typical word / phrase representing the contents of the document group as a label and extracting a specific word / phrase that complements the outline as a keyword, the contents of the document group can be displayed more clearly. Made possible.

  The present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the invention. For example, in the above example, the keywords are extracted together with the labels, but only the labels may be extracted. Conversely, only keywords may be extracted. In the above example, scoring is performed on labels and keywords using inclusion relations in terms of phrases, but semantic inclusion relations are analyzed using a dictionary or the like, and the same scoring is performed based on the analysis. You may do it. However, since the semantic inclusion relationship is opposite to the morphological inclusion relationship, it is necessary to adjust the score in reverse.

It is a block diagram which shows Example 1 of this invention. It is a flowchart of label extraction score calculation in Example 1 of this invention. It is a flowchart of the keyword extraction score calculation in Example 1 of this invention. It is an example of the important phrase and phrase importance score in Example 1 of this invention. It is an example of the table which shows the inclusion relation of the important word phrase in Example 1 of this invention. It is explanatory drawing of an example of the label extraction score in Example 1 of this invention. It is explanatory drawing of an example of the keyword extraction score in Example 1 of this invention. It is explanatory drawing of an example of the label extraction score and keyword extraction score in Example 1 of this invention. It is explanatory drawing of an example of the display output method in Example 1 of this invention. It is explanatory drawing of an example of the display output method in Example 1 of this invention. It is explanatory drawing of an example of the display output method in Example 1 of this invention. It is a block diagram which shows Example 2 of this invention. It is an example of the important phrase and phrase importance score in Example 2 of this invention. It is an example of the table which shows the inclusion relation of the important word phrase in Example 2 of this invention. It is explanatory drawing of an example of the label extraction score in Example 2 of this invention. It is explanatory drawing of an example of the keyword extraction score in Example 2 of this invention. It is explanatory drawing of an example of the label extraction score and keyword extraction score in Example 2 of this invention. It is an example of the important phrase and phrase importance score in Example 2 of this invention. It is an example of the table which shows the inclusion relationship of the important word phrase in Example 2 of this invention. It is explanatory drawing of an example of the label extraction score in Example 2 of this invention. It is explanatory drawing of an example of the keyword extraction score in Example 2 of this invention. It is explanatory drawing of an example of the label extraction score and keyword extraction score in Example 2 of this invention. It is explanatory drawing of an example of the display output method in Example 2 of this invention. It is explanatory drawing of an example of the display output method in Example 2 of this invention. It is explanatory drawing of an example of the display output method in Example 2 of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Phrase extraction part 2 ... Phrase importance score calculation part 3 ... Inclusion relation analysis part 4 ... Label extraction score calculation part 5 ... Keyword extraction score calculation part 6 ... Label selection part 7 ... Keyword selection unit 8 ... Display output unit 9 ... Label / keyword holding unit

Claims (14)

In a document processing apparatus for extracting a label representing the contents of a document set in one word and one or more keywords representing an outline of the document set,
Word / phrase extracting means for extracting words / phrases that are candidates for labels and keywords from each document in the document set;
A phrase importance score calculating means for calculating a phrase importance score representing the importance of the phrase for each phrase;
Inclusion relation analysis means for analyzing a morphological inclusion relation between the words;
Label extraction score calculation means for calculating a label extraction score for each word from the word importance score based on an analysis result of whether or not the word is included in another word analyzed by the inclusion relation analysis means When,
Keyword extraction score calculation means for calculating a keyword extraction score for each of the phrases from the phrase importance score, based on an analysis result of whether or not the phrase includes other words analyzed by the inclusion relation analysis means; ,
Label selection means for selecting one label from the words according to the label extraction score;
A document processing apparatus comprising: a keyword selecting unit that selects a keyword based on the keyword extraction score.   The document processing apparatus according to claim 1, wherein the inclusion relation analyzing unit analyzes the inclusion relation of character strings.   The document processing apparatus according to claim 1, wherein the inclusion relation analyzing unit analyzes the inclusion relation of the word string.   The document processing apparatus according to claim 1, wherein the keyword selection unit does not select an included phrase as a keyword.   5. The document processing according to claim 1, further comprising display means for displaying the label and the keyword, wherein the display means displays the part corresponding to the label in the keyword separately from the other parts. apparatus. The document group classification means for classifying the document group into a plurality of document sets, and the label and the keyword are extracted for each document group classified by the document group classification means. A document classification device comprising the document processing device described above,
The word classification importance calculation means and the label extraction score calculation means calculate a score so that a score of a word selected as a label and a keyword of another document set becomes small. In a document processing apparatus that extracts a label representing the contents of a document set in one word,
A word / phrase extracting means for extracting a word / phrase as a label candidate from each document in the document set;
A phrase importance score calculating means for calculating a phrase importance score representing the importance of the phrase for each phrase;
Inclusion relation analysis means for analyzing inclusion relations between the words;
Label extraction score calculation means for adjusting the word importance score based on the inclusion relation analyzed by the inclusion relation analysis means and calculating a label extraction score;
A document processing apparatus comprising: a label selecting unit that selects a label based on the label extraction score.   The document processing apparatus according to claim 7, wherein the inclusion relationship between the phrases is an inclusion relationship in the form of a phrase.   The document processing apparatus according to claim 7, wherein the inclusion relationship between the phrases is a semantic inclusion relationship of the phrases. In a document processing apparatus that extracts keywords from a document set,
Word / phrase extracting means for extracting word / phrase candidates from each document in the document set;
A phrase importance score calculating means for calculating a phrase importance score representing the importance of the phrase for each phrase;
Inclusion relation analysis means for analyzing inclusion relations between the words;
A keyword extraction score calculating means for adjusting the phrase importance score based on the inclusion relation analyzed by the inclusion relation analyzing means and calculating a keyword extraction score;
A document processing apparatus comprising: a keyword selecting unit that selects a keyword based on the keyword extraction score. In a document processing method for extracting a label representing the contents of a document set in one word and one or more keywords representing an outline of the document set,
A phrase extracting unit extracting a candidate for a label and a keyword from each document in the document set;
A phrase importance score calculation unit calculating a phrase importance score representing the importance of the phrase for each phrase;
An inclusion relationship analysis unit analyzing a morphological inclusion relationship between the words;
A label extraction score calculation unit calculates a label extraction score for each word from the word importance score based on an analysis result of whether or not the word is included in another word analyzed by the inclusion relation analysis unit. A calculating step;
The keyword extraction score calculation unit calculates a keyword extraction score for each word from the word importance score based on the analysis result of whether or not the word includes other words analyzed by the inclusion relation analysis unit And steps to
A label selecting unit selecting one label from the words according to the label extraction score;
A document selection method comprising: a keyword selection unit selecting a keyword based on the keyword extraction score. In a document processing method for extracting a label expressing the contents of a document set in one word,
A phrase extracting unit extracting a candidate for a label from each document in the document set; and
A phrase importance score calculation unit calculating a phrase importance score representing the importance of the phrase for each phrase;
An inclusion relationship analysis unit analyzing an inclusion relationship between the words;
A label extraction score calculation unit adjusting the phrase importance score based on the inclusion relation analyzed by the inclusion relation analysis step to calculate a label extraction score;
A document selection method comprising: a label selection unit selecting a label based on the label extraction score. In a computer program for document processing used to extract a label representing the contents of a document set in one word and one or more keywords representing an outline of the document set,
A word extraction step for extracting candidate words for labels and keywords from each document in the document set;
A word importance score calculating step for calculating a word importance score representing the importance of the word for each word;
An inclusion relationship analysis step of analyzing a morphological inclusion relationship between the words;
A label extraction score calculation step for calculating a label extraction score for each word from the word importance score based on an analysis result indicating whether the word is included in another word analyzed by the inclusion relation analysis step When,
A keyword extraction score calculating step for calculating a keyword extraction score for each of the phrases from the phrase importance score based on an analysis result of whether or not the phrase includes other words analyzed by the inclusion relation analyzing step; ,
A label selection step of selecting one label from the words according to the label extraction score;
A computer program for document processing, which is used for causing a computer to execute a keyword selection step of selecting a keyword based on the keyword extraction score. In a computer program for document processing used to extract a label that represents the contents of a document set in one word,
A word extraction step for extracting word candidates for labels from each document in the document set;
A word importance score calculating step for calculating a word importance score representing the importance of the word for each word;
An inclusion relationship analysis step of analyzing an inclusion relationship between the phrases;
A label extraction score calculation step of adjusting the word importance score based on the inclusion relationship analyzed by the inclusion relationship analysis step to calculate a label extraction score;
A computer program for document processing, which is used for causing a computer to execute a label selection step of selecting a label based on the label extraction score.

Images