JP2002117027A - Feeling information extracting method and recording medium for feeling information extracting program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently extract feeling information from an input text by increasing the precision of the matching of a pattern for extracting the feeling information. SOLUTION: A feeling expression analyzing means 5 generates a feeling expression pattern corresponding to a word or word array having the meaning of a feeling expression from a feeling expression table 6 wherein the correspondence between kinds of feeling information and descriptions of information regarding words expressing the feeling information is recorded. A text analyzing means 3 analyzes and divides an inputted text into words. A feeling information extracting means 7 matches the feeling expression pattern against the word or word string constituting the inputted text and outputs the information which has the kind of the feeling information to correspond to the part of the matched word or word string in the text when the feeling expression pattern matches the word or word string in the text.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for processing a computerized natural language text by a computer, and more particularly, to a method in which emotional expressions such as emotions and emotions to be extracted from the computerized text are defined in advance. In addition, the present invention relates to an emotion information extraction method for automatically adding emotion information as a tag to a partial character string expressing the emotion of an input text.

[0002]

2. Description of the Related Art Conventionally, as a method of extracting emotion information such as emotions, emotions, and so on from a text, a method has been realized in which, when a registered character string or keyword is matched, corresponding emotion information is added. .

[0003]

However, in the above-described conventional emotion information extraction method, all keywords to which emotion information is added must be registered in advance, and words used in words such as declinable words are not registered. There was a problem that it was necessary to register each utilization form. Also,
Since the position in the text is not taken into account, there is a problem that an emotional expression that needs to be taken into account may be erroneously extracted.

The present invention has been made in view of the above points, and uses a regular expression using reading, part-of-speech, or a semantic category in addition to the word notation, instead of matching the keyword expressing the emotion by the word notation. By performing matching, it is possible to cover a large number of expressions derived by registering one emotional expression, and to specify the position using regular expressions. The purpose is to be able to extract emotion information well.

[0005]

SUMMARY OF THE INVENTION In order to solve the above problems, the present invention provides an input means for inputting digitized text, a text analysis means for analyzing the input text and dividing it into words, Emotion expression analysis means for obtaining an emotion expression pattern corresponding to a word or a word string having the meaning of the emotion expression from an emotion expression table in which a correspondence between a type of information and a description of information relating to a word expressing emotion information is recorded; , Matching the emotional expression pattern with the words or word strings that make up the input text, and if the emotional expression pattern matches a word or word string in the text, the part of the word or word string in the matched text Using an emotion information extraction unit that outputs information that associates the type of emotion information that the emotion expression pattern has with the electronic expression input to the computer. Analyzing the text, automatically extracts the emotion information that exists in the text, to give the tag to the word or string of words is included emotional information.

The input means performs a morphological analysis on the input text (see "Natural Language Processing", Makoto Nagao, Iwanami Shoten), divides the text for each word, and writes and reads the divided words. Get part of speech and semantic category numbers.

[0007] The emotion expression analyzing means converts, for example, a semantic category number and a semantic category name from a format described using notations and readings of words, parts of speech, and attributes of semantic category names for each emotion information such as emotions and sorrows. After obtaining a semantic category number obtained by searching for a semantic category name from the paired semantic category table, an expression pattern corresponding to the emotion information is created.

[0008] The emotion information extracting means is for expressing the emotion information using the word notation, reading, part of speech, and semantic category number with respect to the notation and reading of the word string obtained from the text, the part of speech, and the semantic category number. Matching with the pattern is performed, and if the pattern matches, the emotion information corresponding to the pattern is added as a tag only to the word string to which the pattern corresponds, and the text with the tag is output.

A program for realizing the above processing by a computer can be stored in an appropriate recording medium such as a computer-readable portable medium memory, a semiconductor memory, and a hard disk.

[0010]

FIG. 1 is a block diagram showing an embodiment of the present invention. Emotion information extraction processing device 10 shown in FIG.
Is realized by a computer including a CPU and a memory, a software program, and data such as a table and a dictionary stored in a storage device. The input text 11 is
The tagged text 12 is a digitized text from which emotion information is to be extracted, and the type of the extracted emotion information is added to a corresponding word or word string as a tag.
This is the output text.

[0011] The input means 2 is an input text 1 to be processed.
Read 1 The text analysis unit 3 performs a morphological analysis on the input text using the morphological analysis dictionary 4 to divide the text into words, and at the same time, obtains the notation, reading, part of speech, and meaning category number of each word.

As shown in FIG. 2, for example, the morphological analysis dictionary 4 stores, for each morpheme, its notation, part of speech (including part of speech classification), reading, and semantic category number information. It is an electronic dictionary to have. The semantic category number is a numerical value corresponding to a semantic category name defined in advance for the meaning of a morpheme word. Here, the semantic category name is the name of a category in which all words are systematically classified into semantic categories.

The emotion expression analysis means 5 is described as a regular expression using notation and reading of a word string, part of speech, and a name of a semantic category, and stores expressions defined for each emotion information, such as emotions and emotions. The emotion expression table 6 is analyzed to create an expression pattern for each emotion information. At this time, the semantic category number is obtained from the semantic category name by searching the semantic category table 8 in which the semantic category number and the semantic category name are paired. FIG. 3 shows an example of the semantic category table 8.

Expression patterns are sequentially stored in a memory using the length of the word string as a key. The order in which the patterns are stored in the memory may be the order in which the number of characters is small or the order in which they appear in the emotion expression table 6. Alternatively, priorities may be assigned to the emotions in advance, and the feelings may be prioritized in order. What is a regular expression?
This is a standard description method used for pattern matching of character strings.

The emotion information extracting means 7 includes a notation and reading obtained by the text analyzing means 3, a word string accompanied by information of part of speech and a semantic category number, and a notation and reading of the word string obtained by the emotion expression analyzing means 5. The pattern matching is performed with the expression pattern of the emotion information described by the regular expression using the POS, the part of speech, and the semantic category number.

In pattern matching, pattern matching is performed for each of the patterns stored in the memory in the order in which they are stored in the longest match. As a result of the pattern matching, emotion information to which the pattern belongs is added as a tag to the word string portion of the matched text. Here, the longest match means that when there are a plurality of matching candidates, a matching one with the longest character string is selected. For example, consider a case where “not pleasant” exists in the text, and “fun” and “not pleasant” exist in the pattern. In this case, the longest match selects the one that matches with the longest character string.
Instead, "not pleasant" will be selected as the match.

The output means 9 outputs a tagged text 12 to which emotion information is tagged. The control means 1 controls each of these means.

FIG. 4 is a flowchart showing an embodiment of the present invention. Hereinafter, the processing will be described in detail according to steps (a) to (i) shown in FIG.

(A) The sentence expression table 6 which stores the expression of the emotion in the form described by the regular expression is read using the notation, reading, part of speech, and the name of the meaning category of the word string, and the name of the semantic category in the expression is read. The semantic category table 8 is searched by using as a search key to obtain a semantic category number, a pattern of expression of each emotion information is generated, and the pattern is stored in the memory in the long order of the word string.

(B) As long as there is text to be extracted and tagged with emotion information, the following steps (c) to (i)
repeat.

(C) Input text 11 from which emotion information is to be extracted.

(D) The text is divided into sentences, and the sentence is further divided into words by morphological analysis, and the notation, reading, part of speech, and semantic category number of each word constituting the text are obtained.

(E) For each pattern of emotional expression,
The following steps (f) to (h) are repeated.

(F) For all sentences constituting the text, the notation of the word or word string, the reading, the part of speech, and the attribute of the semantic category number are used, and the notation of the pattern is the same as the notation of the word in the text. , The reading of the pattern is compared with the reading of the word in the text, the part of speech of the pattern is compared with the part of speech of the word in the text, and the semantic category number of the pattern is compared with the semantic category number of the word in the text. Check if it matches.

(G) If the pattern matches, the next step (h) is executed.

(H) Only the word or word string corresponding to the pattern is provided with emotion information corresponding to the pattern as a tag.

(I) After the pattern matching for all patterns is completed, a text with a tag is output.

FIG. 5 shows an example of the emotion expression table 6. In the attribute description of FIG. 5, for example, {X | Y | Z}
Indicates that it may be either X or Y or Z.

FIG. 6 shows an example of generating an emotion expression pattern. In FIG. 6, an emotion expression pattern is generated for the notation of the attribute of “anger” in the emotion expression table 6 shown in FIG. 5 by morphological analysis in consideration of the meaning of the symbol of the regular expression. First, since the description of the attribute of “notation = addition / deduction | letter | ro” as shown in FIG. 6A includes the logical sum of the regular expression, FIG.
As shown in FIG. 7, a pattern including three notations of "do not change", "do not change", and "do not change" is generated. Further, each of these notations is divided into words by morphological analysis, and a pattern as shown in FIG. 6C is generated. Each notation in this emotion expression pattern is used to check whether it matches the notation of the word string in the text.

As shown in this example, only the necessary minimum information is registered in the emotion expression table 6 and is expanded into the emotion expression pattern by the emotion expression analysis means 5 to register the emotion expression pattern. Can be easily performed. In particular, when it is not known in advance how the contents of the notation attributes registered in the emotion expression table 6 will be divided into words by the morphological analysis, the emotion expression analyzing means 5 performs word division by the same morphological analysis as the input text 11, that is, Since the conversion to the regular expression is automatically performed, the accuracy of the matching is improved.

In the example of the emotion information “exclamation” in the emotion expression table 6 shown in FIG. 5, both the part of speech and the notation are described as attributes. Is generated.

For example, suppose that there is a sentence in the input text 11 that says, "Never hit me." By the text analysis means 3, the sentence "Never hit me" is divided into the following words. ============================ [Notation] [Reading] [Part of speech] [Semantic category number] Masaka Masaka Combined phrase myself Noun: Pronoun 32 is a ga case particle Ata verb stem 1208 RU Le verb suffix Nante Quote particle N Ne Final particle ============================ === In the matching in the emotion information extraction means 7, the pattern generated by the emotion expression analysis means 5 and the analysis result by the text analysis means 3 are referred to, and these notations, readings, parts of speech, and semantic categories are respectively classified for each word. Compare. The emotional expression pattern of “exclamation” is “part of speech = verb stem, part of speech = verb suffix, notation = ”And“ part of speech = verb stem, part of speech = verb suffix,
Notation = what, and the latter part matches the above-mentioned text “hit (part of speech = verb stem), ru (part of speech = verb suffix), what (notation = what)”.

As shown in this example, the grammatical positions of the main words expressing the emotions are given to the emotional expression patterns by giving attributes such as notation, reading, part of speech or semantic category to each of the word sequences. Can also be specified, so that it is possible to prevent omission of registration of keywords or erroneous matching.

In the example of emotion information of "joy", a symbol and an alphabet are given as notation attributes.
Each is divided into words, that is, symbols and alphabets, and matches when the same sequence of word strings exists in the text.

In the example of emotion information of "surprise",
The attribute of the meaning category name of "surprise" is described. Therefore, a match occurs when a word having a semantic category number of “surprise” exists in the text. For example, if the meaning category number of the word “surprise” is “1”
265 ", which indicates the name of the meaning category" surprise ", the word" surprise "matches the" surprise "pattern.

In the example of the emotion information "call", since the attribute of the notation is accompanied by a regular expression indicating the head of the sentence, it matches a sentence having the expression "by" at the beginning of the sentence.

FIG. 7 shows an example of the input text 11. A tag of emotion information is added to the example of the input text of FIG. 7 according to the emotion expression table 6 of FIG. 5, and the output result is shown in FIG. As is clear from FIG. 8, in the tagged text 12, the tag of the emotion information of “call” is “where”, the tag of the emotion information of “surprise” is “surprise”, and the tag of the emotion information is “＼ ＾ o ＾ /”. Tag of emotion information of "joy",
A tag of emotion information of "exclamation" is given to "hit", and a tag of emotion information of "anger" is given to "let's get rid of". The output of the emotion information extracted according to the present invention is not necessarily limited to the output in the tag format, but can be output in another format.

A program for causing a computer to execute the processing shown in FIG. 4 can be installed in the computer through a CD-ROM or other recording medium or a communication line.

The features and variations of this embodiment will be listed below. (1) The emotion information extraction method according to the present embodiment includes an input process of inputting digitized text, a text analysis process of analyzing the input text and dividing it into words, a type of emotion information and emotion information. Analyzes the emotion expression table in which the correspondence with the description of the information about the word expressing the word is registered, and performs the emotion expression analysis process of generating the emotion expression pattern, and the matching of the emotion expression pattern with the words or word strings that constitute the text. And determining whether or not the pattern matches the emotional expression pattern. If the pattern is matched, the emotional information providing process of providing the type of the emotional information corresponding to the pattern to the word or word string in the matched text. , An output process that outputs text to which emotion information types are assigned, and a control process that controls these processes Ri, analyzes the electronic text entered into the computer, automatically extracts the emotion information is present in the text, to assign tags to the words or word strings includes emotion information. (2) In the above (1), a text analysis process having a function of dividing a text into sentences, dividing the words into words constituting the sentence, and obtaining respective attributes of word description, reading, part of speech, and semantic category number. Having. (3) In the above (1), the expression of the emotional expression is divided into words from the emotional expression described with the attribute of the expression including not only one word but also a plurality of words, and the expression of each word is expressed. It has an emotion expression analysis process with the function of obtaining information and creating patterns for emotion expression. (4) In the above (1), there is an emotion expression analysis process having a function of creating a pattern for an emotion expression from an emotion expression described with an attribute of reading a word. (5) In the above (1), there is an emotion expression analysis process having a function of creating a pattern for emotion expression from an emotion expression described with the attribute of the part of speech of a word. (6) In the above (1), a sentence expression described with an attribute of a semantic category name of a word is used in a text by using a semantic category table in which a semantic category number and a semantic category name of a word are paired. Has a semantic category number from the semantic category name of the word, and has an emotion expression analysis process having a function of creating a pattern for emotion expression. (7) In the above (1), a function is provided for creating a pattern for emotion expression from an emotion expression described in a regular expression using word notation, reading, part of speech, or a meaning category name in the emotion expression table. It has an emotion expression analysis process. (8) In the above (1), a pattern for expressing an emotion including a word notation, a reading, a part of speech, and a semantic category name,
It has an emotion information extraction process having a function of performing pattern matching with text divided into words having attributes of notation, reading, part of speech, and meaning category number. (9) In the above (1), there is an emotion information extraction process having a function of matching with a pattern of emotion information. (10) In the above (1), a word or word string is divided into a pattern for emotional expression consisting of notation and reading, part of speech and a semantic category number, and a word having attributes of notation, reading, part of speech and a semantic category number. When performing pattern matching with a given text, the notation of a pattern word or word string is the word or word string in the text, the reading of the pattern word is the reading of a word in the text, and the word class of the pattern word is It has an emotion information extraction process having a function of comparing the part of speech of a word in the text and the semantic category number of the word of the pattern with the semantic category number of the word in the text. (11) In the above (1), there is an emotion information extraction process having a function of performing pattern matching with the longest match when performing pattern matching between a notation for emotion expression and a word obtained by dividing a text. (12) In the above (1), when the pattern for emotion expression matches the text divided into words, the pattern corresponds to only the partial word string of the text corresponding to the word constituting the pattern. It has an emotion information extraction process provided with a function of giving the type of emotion expression as a tag.

[0040]

As described above, according to the present invention, when an expression in which emotion information is described in a text is automatically extracted, a regular expression using reading, part-of-speech, or a semantic category in addition to word notation is used. Matching can be performed, so that a large number of derived expressions can be covered by registering one emotional expression, and the position can be specified by a regular expression, so that there is no omission of keyword registration or erroneous matching In addition, it is possible to efficiently extract emotion information from a text and attach a tag of the emotion information to a corresponding word or word string in the text.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a morphological analysis dictionary.

FIG. 3 is a diagram illustrating an example of a semantic category table.

FIG. 4 is a flowchart showing an embodiment of the present invention.

FIG. 5 is a diagram showing an example of an emotion expression table according to the embodiment of the present invention.

FIG. 6 is a diagram illustrating an example of generating an emotion expression pattern according to the embodiment of the present invention.

FIG. 7 is a diagram illustrating an example of an input text.

FIG. 8 is a diagram showing an example of a tagged text to be output.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 control means 2 input means 3 text analysis means 4 morphological analysis dictionary 5 emotion expression analysis means 6 emotion expression table 7 emotion information extraction means 8 semantic category table 9 output means 10 emotion information extraction processing device 11 input text 12 tagged text

Claims (5)

[Claims] In a method for extracting emotion information from an expression described in an electronic text, an emotion expression in which a correspondence between a type of emotion information and a description of information on a word expressing the emotion information is recorded. An emotion expression analysis process of obtaining an emotion expression pattern corresponding to a word or a word string having the meaning of the emotion expression from a table, a text analysis process of analyzing input text and dividing the text into words, If the emotion expression pattern matches a word or word string in the text, and the emotion expression pattern matches the word or word string in the text, the emotion expression is added to the word or word string part in the matched text. An emotion information extraction step of outputting information corresponding to the type of emotion information of the pattern. Method. 2. The emotion information extraction method according to claim 1, wherein the type of the emotion information extracted in the emotion information extraction step is added as a tag to a word or a word string in the matched text. Emotion information extraction method. 3. The emotion information extraction method according to claim 1, wherein the emotion expression table includes one or more words as a description of information relating to a word expressing emotion information corresponding to a type of emotion information. An emotion information extraction method characterized by having notation, reading, and description of part of speech or semantic category information of emotion expressions composed of words. 4. The emotion information extracting method according to claim 1, wherein in the text analysis step, the sentence of the input text is divided into words, and each sentence is written and read. , Part-of-speech or semantic category information, and the emotional expression pattern is provided with notation, reading, part-of-speech or semantic category information, or any one of these for words or word strings constituting the pattern. Have multiple combinations of attributes,
In the emotion information extracting step, matching is performed for each attribute of the emotion expression pattern and a word or word string in a text. 5. A recording medium on which a program for extracting emotion information from a description described in an electronic text by a computer is recorded, the information being related to a type of emotion information and a word expressing emotion information. Expression analysis processing for obtaining an emotion expression pattern corresponding to a word or a word string having the meaning of the emotion expression from an emotion expression table in which correspondence with the description of the expression is recorded, and analyzing the input text to divide it into words. And performing matching between the emotion expression pattern and the words or word strings constituting the input text. When the emotion expression pattern matches a word or word string in the text, the matching Emotion information extraction that outputs information that associates the type of emotion information of the emotion expression pattern with the word or word string part of A recording medium for an emotion information extraction program, wherein a program for causing a computer to execute output processing is recorded.