JP2002366557A - Retrieval device, computer program and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To exactly extract a sentence including objects such as words and phrases to conform to retrieval conditions from a document. SOLUTION: This retrieval device is provided with a text dividing means 1 to divide an English text document 100 to be retrieved by every sentence, a retrieval condition accepting means 2 to accept part or the entire part of the objects to be extracted, information in relation to parts of speech such a names, conjugation forms of parts of speech of the objects or only a piece of information in relation to the parts of speech as the retrieval conditions, lists 5, 6 in which original forms and irregular change forms of words are registered by every part of speech such as nouns, verbs and adjectives, a retrieving means 3 to retrieve the document to be retrieved by referring to the lists 5, 6 and a retrieval result presenting means 4 to extract and present the sentence including the objects conforming to the retrieval conditions.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention
Extract sentences that contain the object that matches the search criteria,
The present invention relates to a search device that presents a sentence of a search result including an object, a computer program for realizing the device with a computer, and a recording medium on which the computer program is recorded.

[0002]

2. Description of the Related Art As a function of searching for a word in a document, document creation and editing software such as word processing software searches the document and jumps to a position of a word that matches a keyword specified in advance. There is a search function.
At this time, as a method of designating the search condition, there is a method of designating only a part of the keyword character string in addition to a method of designating the entire character string of the keyword.

As another example, there is a search engine that presents a list of URLs, outlines, and the like of pages including a keyword specified in advance from a huge number of homepages published on the Internet to a user. . In this search engine, it is possible to set multiple keywords. In addition to keywords, AND conditions, OR conditions, refinement, etc., settings of pages that include both of these keywords, information of pages that include any of these keywords Is displayed in a list.

[0004]

As described above, according to the conventional search function, a user can jump to a position where a keyword appears in a document, or search for a page (file) containing the keyword.
Although it displays a list of URLs and summaries, it does not provide a function for extracting and presenting all sentences in which keywords appear. Such a search function is useful for, for example, accumulating example data for grammar research, analyzing the use tendency of a term in a document such as a novel, the style of a writer, and the like.

[0005] In addition, in the conventional search function, for example, in order to extract only the highest adjective from an English sentence, "* e
When the keyword "st" (* is an arbitrary character string) is set, not only the highest level of adjectives but also words that are not searched, such as nouns, are extracted, so that the search efficiency is low.

Further, although a function of displaying a list of information such as a URL and an outline of a home page including all of a plurality of keywords is provided, the order of appearance of the plurality of keywords cannot be specified. Also, when trying to search for attribute information such as part of speech other than a character string, inflected forms, etc. as a keyword, it is possible to set up to a certain extent by applying a format called a regular expression method that uses more than 10 symbols, It is quite difficult to cover irregular usage. Also, learning the regular expression method requires deep knowledge and experience. Further, the conditional sentence set by the regular expression method is difficult to understand by anyone other than the expert, and the visibility is poor.

The present invention has been made in order to solve such a problem. A list in which words are registered for each part of speech is provided, a document to be searched is divided into sentence units, and not only a character string but also a character string is used. So that information related to the part of speech, such as part of speech, inflected forms, etc., can be combined or set as a search condition. In the case of a search condition consisting of multiple items, the order of appearance of the items, the number of words not searched Or a search device capable of accurately extracting sentences including object words such as words, collocations, and correlated words that match the search conditions from a document by enabling setting of impurity conditions such as presence or absence as search conditions. It is an object of the present invention to provide a computer program for realizing the apparatus by a computer, and a recording medium on which the computer program is recorded.

[0008]

According to a first aspect of the present invention, there is provided a retrieval apparatus comprising:
In a retrieval apparatus for retrieving documents and extracting object words that match search conditions, a list in which words are registered by part of speech, means for separating documents into sentences, a part or all of the object words, and / or Or means for receiving information relating to the part of speech of the object as a search condition, and means for searching a document by referring to the list, extracting and presenting a sentence including the object word that matches the search condition. It is characterized by.

[0009] In addition to the first invention, the search device of the second invention is characterized by further comprising means for associating a sentence of the presented search result with a location of the document where the sentence appears, and presenting the sentence. .

[0010] In the search device of the third invention, the object of the first or second invention is a collocation consisting of a combination of a plurality of continuous words, and a part or all of each of the plurality of words, and / or Alternatively, in addition to the information relating to the parts of speech of the plurality of words, information defining the order of appearance of the plurality of words is received as a search condition.

[0011] A search device according to a fourth invention is characterized in that the search device according to the third invention is adapted to accept, as a search condition, information defining the number or presence / absence of a non-search target word intermingled between the plurality of words. I do.

A search device according to a fifth aspect of the present invention is the search device according to any one of the first to fourth aspects, wherein the list of the parts of speech in which the form of the word changes in accordance with the grammar includes the list of the original form of the word and the list of the irregular change. If the target word that is included in the list of the original and irregular variations does not exist as a target for determining whether or not it matches the search condition including the information related to the part of speech, the target word is a regular one. Means for determining whether or not the original form of the original character string is determined. It is characterized in that it is determined whether or not the target word is a search target object based on whether or not a column exists in the list.

[0013] A computer program according to a sixth aspect of the present invention is a computer program used in a computer that searches a document by referring to a list in which words are registered by part of speech and extracts object words that match search conditions. Means for dividing a document into sentences, means for receiving part or all of the object and / or information relating to the part of speech of the object as a search condition, and searching for the document with reference to the list. It is characterized by functioning as a means for extracting and presenting a sentence containing an object matching the condition.

According to a seventh aspect of the present invention, in addition to the sixth aspect, the computer program causes the computer to function as means for presenting a sentence of the presented search result in association with a location of the document where the sentence appears. Features.

According to an eighth aspect of the present invention, in the computer program according to the sixth or seventh aspect, when the object is a collocation consisting of a combination of a plurality of continuous words, the computer program causes the computer to execute a part or a part of each of the plurality of words. The present invention is characterized in that it functions as a means for receiving, as a search condition, information defining the order of appearance of the plurality of words, in addition to all and / or information related to the parts of speech of the plurality of words.

A computer program according to a ninth invention causes the computer to function as means for accepting, as a search condition, information defining the number or presence / absence of non-search words intermingled between the plurality of words of the eighth invention. It is characterized by.

A computer program according to a tenth aspect of the present invention is the computer program according to any one of the sixth to ninth aspects, wherein the list of the parts of speech in which the form of the word changes according to the grammar includes the list of the original form of the word and the list of the irregular form. If the term is not included in the list of the original form and the irregular variation form, the computer is re-evaluated. Means for determining whether or not the word is a regular variation; means for inverting the original character string from the target word if it is determined to be a regular variation; and It is characterized by functioning as a means for determining whether or not the target word is a search target object based on whether or not a character string exists in the list.

The recording medium according to the eleventh aspect of the present invention stores a computer program for causing a computer to search for a document by referring to a list in which words are registered for each part of speech and to extract a target word that matches a search condition. In a readable recording medium, a computer has a program code means for dividing a document into sentences, and a computer has a part or all of the object,
And / or program code means for receiving information relating to the part of speech of the object as a search condition, and causing a computer to search for a document by referring to the list and extract a sentence including the object word that matches the search condition. Computer program including a program code means to be presented.

According to the present invention, a document to be searched is divided for each sentence, and a part or all of an object to be extracted and information related to the part of speech such as a part of speech name, an inflected form or the like of the object word, or a part of speech. Only relevant information is accepted as search criteria,
The document is searched by referring to a list in which words are registered for each part of speech such as a noun, a verb, and an adjective, and a sentence including an object word that matches the search condition is extracted and presented.

Therefore, the object to be extracted can be accurately extracted. In addition, the words before and after the object are understood, and are useful for analyzing, for example, how the object is used in the sentence, affinity of close words, tendency of style, and the like.

Further, in the present invention, in response to a click on a sentence of the presented search result, a portion of the document where the sentence of the detected result appears is presented in association with highlighting, color change, or the like.

Therefore, the context of a sentence including an object can be known, which is useful, for example, for analyzing the tendency of style.

In the present invention, when the object is a so-called collocation, part or all of each word, the part of speech name of a plurality of words,
In addition to information relating to parts of speech such as inflected forms or information relating to parts of speech forming a collocation, information defining the order of appearance of a plurality of words is accepted as a search condition.

Therefore, when a plurality of search conditions are set in the order of "a-", "b-", "c-", "b-a-c-", "a-c-b-""B-c-a
~), Or a group of words appearing separately in a plurality of sentences within a sentence, such as a sentence such as
An accurate search result of only the collocations in the order of appearance such as "bc" is obtained.

According to the present invention, when the above-described search condition is set, information defining the number or the presence or absence of a non-search target word intermingled between a plurality of words is accepted as the search condition.

Therefore, although a plurality of words are used in combination, such as "so-that", collocations in which an arbitrary word is interposed therebetween can be accurately extracted.

Also, in the present invention, the third person singular present tense, past tense, past participle, and ing of verbs and auxiliary verbs, comparative classes of adjectives and adverbs, superlative forms, and inflected forms of personal pronouns, etc.
In the case of a part-of-speech having a regular or irregular variation other than the original form, only the word of the irregular variation is registered in the list, and for example, in the case of a verb, "s / es / ed / in
g ”or adjectives and adverbs ending with“ er / e ”
It is determined whether it is a regular variation based on the existence of `` st '', and if it is determined that it is a regular variation, the original character string with these endings removed is reversely generated, and the reverse Based on whether the generated character string exists in the original list, it is determined whether or not the object satisfies the search condition.

Therefore, it is not necessary to register regularly changing words in the list in accordance with the usage, and the memory can be effectively used by reducing the storage amount.

[0029]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG.
FIG. 2 is a block diagram of the present invention device. The text dividing means 1 divides an English text document 100 such as a novel which is publicly available on the Internet into sentence units by dividing the document at a position where a period, a question mark, or an exclamation mark appears, and identifies each sentence. And save it with a serial number etc.

The search condition accepting means 2 provides a layout condition setting screen (see FIG. 2) as described later as a user interface, and displays a character string entered by a user in each column in a condition setting frame or a pop-up list. Part-of-speech names selected from among them, inflected forms (inflected forms), Groups (types of intransitive verbs, transitive verbs, etc. or file names of word lists in which the user has registered any words in advance), and multiple conditions by selecting radio buttons Whether or not the appearance order of the items is specified, the number of the contaminating words in the collocation using a symbol such as “*” defined as an arbitrary word, and the number of the contaminating words obtained by combining this symbol with a number are received as search conditions.

Based on the set search conditions, the search means 3 includes a part-of-speech word list (original form) 5 in which words are registered for each part of speech such as a noun, an intransitive verb, a transitive verb, an adjective, etc. The part-of-speech word list (irregular variant) 6, which is registered for each part-of-speech such as noun, transitive verb, intransitive verb, adjective comparative grade, superlative adjective, inflectional form of pronoun, etc. It describes the three basic forms of the auxiliary verb, the past tense, the past participle, the ing form, and the regular use forms of the adjectives and adverbs, including the comparative grades and superlative forms, and the prosodic inflection forms. The text division unit 1 refers to one of the rule utilization files 7
Retrieves English text data divided into sentence units, and extracts sentences containing object words that match the search conditions.

The search means 3 stores the extracted sentence together with information (for example, a serial number of the sentence, a keyword existing position, etc.) enabling linking with the sentence of the English text data.

In this example, the search result presenting means 4 has a function of presenting up to five types of search results on separate screens as shown in FIG. When the sentence is clicked, the location of the English text document where the sentence of the detection result appears is presented in association with the sentence of the detection result by highlighting, color change, or the like.

FIG. 2 is a layout diagram of a search condition setting screen. In this example, up to eight items (eight columns) can be set for one condition. The contents of the condition include the character string, part-of-speech, verb, and auxiliary verb in triads (third-person singular present tense), past tense, past participle, and ing form, as well as comparisons of adjectives and adverbs, and "er / est"", A variation (inflectional form) such as a personal pronoun's case change, and a Group (a type of intransitive verb, transitive verb, etc., or a file name of a word list in which the user has registered an arbitrary word in advance) can be set. In the setting of the part of speech, a user interface is provided which enables setting by selection from a pop-up list including a list of part of speech names.

Note that the complete word character string that can be specified in one cell is not limited to one word, and for example, "/"
Any number of words separated by can be specified collectively. With such a specification, an OR search of a plurality of words at that position can be specified. For example, in one column, if "take / get / put" is specified in one cell of the character string, "verb" is specified in the part of speech, and "past tense and past participle" are specified in the inflected (conjugated) form, take or The past tense and past participle of get or put are retrieved and extracted.

Regarding the variation (conjugation type), for example,
In the case of a verb, specify the inflected form (variant): 0: original form, 1:
Santangen (= present tense), 2: past tense, 3: past participle, 4: ing
For adjectives and auxiliary verbs, 0: present tense, 1: past / past participle, 2: ing, and for adjectives (adverbs), 0: original,
Provide a user interface that can be set by selecting from a pop-up list consisting of a list of applicable choices, such as 1: comparative grade, 2: superlative.

At this time, if all of the inflected forms are selected as the search conditions, the column of the modified form is blank, and if any of the inflected forms is selected, a number (column) corresponding to the inflected form is used.
Notation.

In addition, in the OR search of a plurality of words, in addition to specifying using the "/" symbol in a cell of a character string as described above, a part-of-speech-based reference file created by a user by pre-registering words. By specifying the file name in the Group cell, it is possible to specify the OR search of multiple words registered in the reference file.

As described above, in the conventional regular expression method, the user has to make full use of complicated symbol groups in setting search conditions, and also requires a complicated token analysis in computer processing. According to the present invention, the search condition can be specified in a two-dimensional array grid format, so that the visibility is excellent, the operation of specifying the condition is much easier, and the load of the condition analysis process in computer processing is reduced.

The designation order / non-designation of the appearance order of the condition items can be set by a radio button of “designation order / out of order”. Others
Ten types of ranges including “all text” can be specified as “extraction source”, and a presentation screen can be specified from ten screens described later as “extraction destination”.

Note that the search condition in FIG.
Unless "is below (set conditions)" or "contains"
"English", "below", "begins with", "English", "below"
It is also possible to specify "end with", "English", "below" or "does not include".

Further, it is possible to make an ambiguous designation without specifying a character in the character string field. “*” Is an arbitrary character string of one or more characters, and a combination method with the characters is defined as follows, for example. "A *": word starting with a "* a": word ending with a "a * b": word starting with a and ending with b "a * b *": word starting with a and containing b "* a * b"": Word including a and ending with b Note that the above-mentioned" a "and" b "indicate arbitrary characters, and the number may be two or more characters.

"*" Is defined as specifying the following contaminant condition when combined with a numeral without specifying the part of speech. When there are two or more specified words and it is desired to specify the interval between two words (the number or presence or absence of foreign words), specify "*" in combination with a number. Here, “n” and “m” represent arbitrary numbers. "*": The interval between two words is free (there is no need to have any foreign words) "* n": Limited to n words between the two words "* nm": Between two words "* -N": limited to words with no or less than n words between two words "* n-": n is a word between two words Must be at least words

Further, by combining "*" with the range of the first character (for example, "xz") and designating it as "xz *", any character string, "x" or "y" or A condition that only character strings starting with “z” are searched and extracted can be specified.

Two sets of the above "conditions" can be set, and an AND condition or an OR condition can be specified between the two sets of conditions. “History” of search conditions set in the past is stored for a predetermined number.

FIG. 3 is a layout diagram of a search result presentation screen. In the figure, the function of arrow A is a function of displaying collocation of search results. This function enables sorting of keywords and arbitrary words before and after the keywords. By aligning the keywords vertically, it is possible to analyze the collocation generation situation.

The function of arrow B is a function of switching a target word to be used as a collocation axis. For example, "a""b"
When a sentence containing an object using the three elements “c” as a search condition is extracted, the axes to be aligned vertically are “a”, “b”, and “c”.
It is a function that can be arbitrarily switched between and can analyze the collocation generation situation for each element.

The arrow C links the extracted sentence and the appearance position in the original sentence. When the sentence (inverted sentence) is clicked in the collocation display (A) of the extraction result, another arrow C is displayed. The corresponding sentence of the original text displayed in the window is highlighted in bold or the like. Thereby, the sentence before and after the extracted sentence can be easily referred to. At this time, the condition element (to buy) in the sentence in the original text is displayed, for example, in red.

The arrow D is a function showing the density distribution in the original sentence of a sentence including words and collocations that match the search condition. Here, the total width corresponds to the total number of sentences in the original text,
The vertical width corresponds to 6 sentences, and if there is a hit sentence, the corresponding position is displayed in red. This function can be applied to writer's style studies.

An arrow E indicates that a function for holding the extraction results in a multi-layered manner for ten faces is provided, and the screen can be switched and displayed by clicking a tab. This function is a function that can further narrow down the extraction results or extract object words from the same original text under different conditions, can hold those results for 10 pages, and compare the search results.

Next, the search procedure of the apparatus of the present invention will be described with reference to the condition setting example shown in FIG. 4 and the flowcharts shown in FIGS.

First, a procedure for selecting a branch destination routine of each column based on the set search conditions will be described with reference to the flowchart of FIG. For example, FIG.
When the search condition is set as shown in (c), first, the character string in the character string column of the first column is determined (S1), and it is determined whether "*" is included in the character string (S2). ). If “*” is not included (in the case of NO), it is determined whether or not a part of speech is specified by referring to the part of speech column of the first column (S3).

When the part of speech is specified as a result of the determination in step S3 (in the case of YES), the "selection routine flag" of the "complete word inspection routine with part of speech attribute" is set for this column (S4). The “complete word inspection routine with part-of-speech attribute” is a routine that extracts a sentence including a word having a designated character string as its original form and including a word that matches the part-of-speech and inflected form designation. For example, this is a routine for executing a procedure as shown in FIG. 6 described below under the setting conditions as shown in FIG.

On the other hand, if the result of determination in step S3 indicates that no part of speech is specified (NO), the "selection routine flag" of the "complete word inspection routine without part of speech attribute" is set for this column (S5). ). The "complete word inspection routine without part-of-speech attribute" is a routine for extracting all sentences including a word that completely matches the specified character string.

If it is determined in step S2 that the character string contains "*" (YES), it is determined whether the character string is only "*" (S6). If the result of determination in step S6 is that the character string is only "*" (in the case of YES), it is determined whether or not a part of speech is specified by referring to the part of speech column (S7).

When the part of speech is specified (in the case of YES),
The “selection routine flag” of the “complete arbitrary word inspection routine with part of speech attribute” is set for this column (S8).
The “complete arbitrary word inspection routine with part of speech attribute” is a routine that extracts all sentences that match the designation of part of speech and inflected forms and include a word whose character string is not specified.
For example, this is a routine for executing a procedure as shown in FIGS. 7 and 8 described below under setting conditions such as the first, second, and third columns in FIG. 4B.

On the other hand, if the result of determination in step S7 indicates that no part of speech is specified (NO), the "selection routine flag" of the "word count-independent foreign word processing routine" is set for this column (S9). ). The “word count-independent foreign word processing routine” is executed when there are two or more specified words and the interval between the two words is free, that is, all sentences including collocations or correlated words regardless of the number and presence of the foreign words. Is a routine for extracting.

As a result of the determination in steps S2 and S6, if the character string contains "*" (in the case of YES), but is not only "*" (in the case of NO), a character string such as "a *" is used. To determine if the character is a specified "ambiguous type"
(S10).

As a result of the determination in step S10, when it is determined that the type is "ambiguous type" (in the case of YES), it is determined whether or not a part of speech is specified by referring to the part of speech column (S11).
If the part of speech is specified, the "selection routine flag" of the "ambiguous word inspection routine with part of speech attribute" is set for this column (S12). The “ambiguous word inspection routine with part-of-speech attribute” is a routine that extracts all sentences that include a specific character and include a word that matches the part-of-speech and inflected form designation. For example, this is a routine for executing a procedure as shown in FIG. 9 described below under setting conditions as in the first column of FIG. 4C.

As a result of the determination in steps S10 and S11, if the type is "ambiguous type" (YES), but no part of speech is specified (NO), the "ambiguous word inspection routine without part of speech attribute""Selection routine flag"
Is set (S13). The “ambiguous word inspection routine without part of speech attribute” is a routine for extracting all sentences including a word including a specific character.

If the result of the determination in step S10 is not "ambiguous type" (in the case of NO), it is determined that the character is a "contaminant word designation type" in which "*" in the character string column is combined with a number. The "selection routine flag" of the "word number designation impurity word processing routine" is set for the column (S14). The "word number designation impurity word processing routine" is a routine for extracting all sentences including a collocation word or a correlated word in which there are two or more designated words and an interval between two words is designated.

Similarly, a routine selection process is performed for each column below the second column, and a “selection routine flag” is set for each column.

FIG. 6 is a flowchart showing a search procedure when all the character strings are specified and the word having the part of speech designation is a search condition (FIG. 4A). In FIG. 4 (a),
If there is no designation in the "Variant" column,
It indicates that all of the past tense, past participle and ing are conditions, and that "part of speech" is "verb" and "type (Group)"
Indicates that both of "intransitive verb and transitive verb" are conditions.

One word at the beginning of the sentence is set as the main target word.
(S401), referring to the "selection routine flag" in the first column, branch to the "complete word inspection routine with part of speech attribute" of the verb, and first determine whether the main target word exists in the intact list of intransitive verbs and transitive verbs Is determined (S402), and if it exists in the verb prototype list, the process proceeds to hit processing (S406).

On the other hand, if the result of the determination in step S402 indicates that the main target word does not exist in the original intransitive verb and transitive verb list, it is determined whether or not it exists in the irregular invariant list of intransitive verbs and transitive verbs (S403). If it exists in the irregular verb variation list, the process proceeds to hit processing (S406).

On the other hand, if the result of the determination in step S403 indicates that the main target word does not exist in the irregular inflection list of the intransitive verb and the transitive verb, whether or not “s / es / ed / ing” is at the end of the main target word Is determined (S404).

If the main target word has "s / es / ed / ing" at the end, the original character string is inversely generated in accordance with the verb rule utilization method (S405), and the process returns to step S402 to generate the inversely generated character string. It is determined whether the prototype character string exists in the verb prototype list. Here, if it exists in the verb prototype list, the process proceeds to hit processing (S406).

If the result of determination in step S404 is that there is no "s / es / ed / ing" at the end of the main target word, or if the original character string inversely generated in step S405 does not exist in the verb original list, a no hit Move to processing (S407).

Hit processing (S406) or no hit processing (S40)
When step 7) is completed, the main target word is set to the next right word (S40).
8) It is determined whether or not the sentence is the last based on whether or not the right word to be set exists (S409).

If it is not the end of the sentence, and if the hit processing is completed, the processing of the routine corresponding to the "selection routine flag" set in the next column is executed for the next main target word. If no hit processing is completed, the process returns to step S402, and the above-described processing of steps S402 to S409 is repeated for the next main target word. On the other hand, if the result of the determination in step S409 is that the sentence is the last sentence (in the case of YES), the target sentence is set to the next sentence (S410), and based on whether or not the next sentence to be set exists. Then, it is determined whether or not it is the end of the document (search range) (S411).

If it is not the end of the document, the process returns to step S401, and the above-described processes of steps S401 to S409 are repeated for each target word of the next sentence. On the other hand, if the result of determination in step S411 is that the document is at the end of the document (search range) (YES), the search process ends.

FIGS. 7 and 8 are flowcharts showing a search procedure in the case where a collocation word which does not specify a character string but only a part of speech is a search condition (FIG. 4 (b)). In this example, the structure of "auxiliary verb + verb + preposition" (for example, "had gone with",
An example will be described in which all sentences including “will abondon on” are extracted.

One word at the beginning of the sentence is set as the main target word.
(S801), referring to the “selection routine flag” in the first column, branching to the auxiliary verb “complete arbitrary word inspection routine with part-of-speech attribute”, first determining whether the main target word exists in the auxiliary verb original list. It is determined (S802), and if it is not present in the auxiliary verb original form list, it is determined whether or not it is present in the irregularly changed auxiliary verb list (S803).

If the main word of interest does not exist in the auxiliary verb original form list or the irregular variation list, the main word of interest is set to the next right word (S809), and the flow returns to step S802 to return to the next main attention word. Start processing word.

If the main word of interest is present in the auxiliary verb original or irregular variation list, reference is made to the "selection routine flag" in the second column to branch to the verb "complete arbitrary word inspection routine with part of speech attribute". Then, first, it is determined whether or not the word to the right of the main target word exists in two types of verb intact lists (intransitive verb and transitive verb) (S804). Move to step S810.

On the other hand, if the result of determination in step S804 is that this word does not exist in either of the two verb intact lists,
It is determined whether or not there are two types of irregularly changed verbs (intransitive verb and transitive verb) (S805). If it is found in one of the two types of irregularly changed verbs, the process proceeds to step S810.

On the other hand, as a result of the determination in step S805, if this word does not exist in any of the two types of irregular verb variations, it is determined whether or not “s / es / ed / ing” is at the end of this word. Is determined (S806).

If the word ends with "s / es / ed / ing", the original character string is inversely generated according to the verb rule usage.
(S807), it is determined whether or not the reversely generated prototype character string exists in the verb prototype list (S808). If it exists in the verb prototype list, the process proceeds to step S810.

As a result of the determination in step S806, if there is no “s / es / ed / ing” at the end of this word, or as a result of the determination in step S808, the original character string reversely generated in step S807 is If it does not exist in the
The next right word is set (S809), and the process returns to step S802 to start processing for the next main target word.

If the main word of interest and the word to the right of it are present in the corresponding part of speech list, the third column is further referred to as “selection routine flag”, and the preposition “complete arbitrary word inspection with part of speech attribute” is performed. Routine ", that is, a routine for determining whether the word to the right is present in the preposition list (S81).
0), if it exists in the preposition list, after hit processing, the next sentence is set as a target (S811), and based on whether or not the target sentence to be set exists, whether the end of the document (search range) is determined. It is determined whether or not it is (S812).

If it is not the end of the document, the flow returns to step S801, and the above-described processing of steps S801 to S810 is repeated for each target word of the next sentence. On the other hand, if the result of determination in step S812 is that the document is at the end of the document (search range) (YES), the search process ends.

As a result of the determination in step S810, if the main target word and the word to the right of it are present in the corresponding part of speech list, but the word to the further right is not present in the preposition list, Based on whether the word following the word to be determined in step S810 is a period, a question mark, or an exclamation mark, it is determined whether the sentence is at the end (S813).

If it is not the last sentence, the main word of interest is set to the next word to the right (S809), and the process returns to step S802 to start the processing for the next main word of interest. Meanwhile, step S8
If the result of the determination in step 13 is the end of the sentence (in the case of YES), after the no hit processing, the next sentence is set as a target (S814), and based on whether or not the target sentence to be set exists. Then, it is determined whether or not the end of the document (search range) (S812).

If the document is not the last, the process returns to step S801, and the above-described processes of steps S802 to S810 are repeated for each target word of the next sentence. On the other hand, if the result of determination in step S812 is that the document is at the end of the document (search range) (YES), the search process ends.

FIG. 9 shows a case where a part of a character string is specified, and a word having a part of speech and a use designation is a search condition (FIG. 4 (c)).
6 is a flowchart showing a search procedure of the search. In this example,
A case will be described as an example in which all sentences including a comparative level or the highest grade of an adjective beginning with “p” are included (that is, the original form of the adjective is excluded).

The first word of the sentence is set as the main target word.
(S1201), it is determined whether or not the main target word is a word starting with “p” (S1202). If the main word of interest is a word beginning with "p", the first column is referred to as "selection routine flag", and branching to the adjective "ambiguous word check routine with part-of-speech attribute" is performed. It is determined whether it exists in the list (S1203). If the main term of interest is present in the adjective prototype list (if present), the search condition is not satisfied, so no-hit processing (S1204) is executed, and the main term of interest is set to the next right word (S1210).

Also, as a result of the determination in step S1203,
If the subject word starting with "p" does not exist in the adjective prototype list (if none), the adjective irregular variant list 2
Determine whether or not it is present in the species (comparative and superlative) (S
1205). On the other hand, the main target word is the adjective irregular variation list 2
Hit processing if present in one or both species
(S1206) to set the main target word to the next right word (S1206).
1210).

If the result of the determination in step S1205 shows that the main word of interest does not exist in either of the two adjective irregular variation lists, it is determined whether the ending of the main word of interest has "er / est". It is determined (S1207), and if there is no `` er / est '' at the end,
The main target word is set to the next right word (S1210).

On the other hand, as a result of the determination in step S1207, if the main target word has “er / est” at the end, the original character string is inversely generated according to the adjective rule utilization method (S1208), and the inversely generated original character It is determined whether or not the column exists in the adjective prototype list (S1209). If it exists in the adjective original form list, hit processing (S1206) is executed, and the main target word is set to the next right word (S1210).

As a result of the determination in step S1209, step S1
If the original character string inversely generated in step 208 does not exist in the adjective prototype list, the main target word is set to the next word to the right
(S1210). If the result of determination in step S1202 is that the main word of interest is not a word starting with "p", the main word of interest is set to the next right word (S1210).

Based on whether the set right word is a period or not, it is determined whether or not the sentence is at the end (S121).
1) If the sentence is not at the end, if the hit processing is completed, the processing of the routine corresponding to the “selection routine flag” set in the next column is executed for the next main target word. If no hit processing has been completed, the process returns to step S1202, and the above-described steps S1202 to S1211 are repeated for the next main target word.

On the other hand, if the result of determination in step S1211 is that the sentence is the last sentence, the target sentence is set to the next sentence (S1212), and based on whether or not the next sentence to be set exists. It is determined whether or not it is the end of the document (search range) (S1213).

If the document is not the last, step S1201
The process of steps S1202 to S1211 is repeated for each target word of the next sentence. on the other hand,
If the result of determination in step S1213 is that the document is at the end of the document (search range) (YES), the search process ends.

In the above description, the case where the search target is an English sentence is described as an example. However, the type of language is not limited to English, and the present invention can be applied to other types of languages.

Further, the procedure of the search processing is not limited to the procedure of the present example, and other procedures can be applied.

The computer program for English sentence retrieval described above can be provided by being preinstalled on a computer, or can be provided on a portable recording medium such as a CD-ROM or MO. Further, it can be provided via a line.

[0097]

As described above, according to the present invention, a list in which words are registered for each part of speech is provided, and documents to be searched are separated into sentence units, and not only character strings but also parts of speech such as parts of speech and inflected forms are used. In order to be able to set related information together or only this information as a search condition, and in the case of a search condition consisting of multiple items, the search conditions are set as the appearance order of items, the number or presence or absence of words not to be searched, etc. So that words that match the search criteria,
This has an excellent effect that sentences including object words such as collocation words and correlated words can be accurately extracted.

[Brief description of the drawings]

FIG. 1 is a block diagram of the device of the present invention.

FIG. 2 is a layout diagram of a search condition setting screen.

FIG. 3 is a layout diagram of a search result presentation screen.

FIG. 4 is a diagram illustrating a setting example of a search condition.

FIG. 5 is a flowchart of a procedure for selecting a branch destination routine based on a set search condition.

FIG. 6 is a flowchart illustrating a search procedure when all of the character strings are specified and a word having a part of speech designation is a search condition.

FIG. 7 is a flowchart (No. 1) illustrating a search procedure in the case where a collocation word that does not specify a character string but only specifies a part of speech is a search condition.

FIG. 8 is a flowchart (part 2) illustrating a search procedure in the case where a collocation word that does not specify a character string but only specifies part of speech is a search condition.

FIG. 9 is a flowchart illustrating a search procedure in which a part of a character string is specified and a word having a part of speech and a use designation is a search condition.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Text division means 2 Search condition acceptance means 3 Search means 4 Search result presentation means 5 Part-of-speech word list (original form) 6 Part-of-speech word list (irregular variation form) 7 Rule utilization file 100 English text document

Claims (11)

[Claims] 1. A retrieval apparatus for retrieving a document and extracting an object that matches a retrieval condition, a list in which words are registered for each part of speech, means for dividing the document into sentences, and a part of the object Or a means for receiving information relating to the part of speech of the object as a search condition, and searching for a document by referring to the list, extracting and presenting a sentence including the object word that matches the search condition And a search device. 2. The retrieval apparatus according to claim 1, further comprising means for associating a sentence of the presented search result with a location of the document where the sentence appears, and presenting the sentence. 3. The object word is a collocation consisting of a combination of a plurality of continuous words, and includes a part or all of each of the plurality of words and / or information related to the part of speech of the plurality of words. 3. The search device according to claim 1, wherein information defining the order of appearance of the plurality of words is received as a search condition. 4. The search device according to claim 3, wherein information defining the number or presence / absence of non-search-target words contaminated between the plurality of words is received as a search condition. 5. The list of parts of speech in which the form of a word changes in accordance with the grammar includes a list of the original form of the word and a list of irregular forms of the word, and determines whether or not the search condition includes information relating to the part of speech. Means for determining whether or not the word of interest is a regular variation if neither the original word nor the list of irregular variations is present; Means for inversely generating a character string of the original form from the word of interest when it is determined to be a variant, based on whether or not the character string of the inversely generated original form exists in the list, The search device according to any one of claims 1 to 4, wherein it is determined whether or not is an object to be searched. 6. A computer program used in a computer for searching a document by referring to a list in which words are registered for each part of speech and extracting an object that matches a search condition. Means for delimiting, part or all of the object, and / or means for receiving information relating to the part of speech of the object as a search condition, and object searching for a document by referring to the list and matching the search condition A computer program for functioning as a means for extracting and presenting a sentence including: 7. The computer program according to claim 6, wherein the computer is caused to function as means for presenting a sentence of the presented search result in association with a location of the document where the sentence appears. 8. When the object is a collocation consisting of a combination of a plurality of consecutive words, the computer associates a part or all of each of the plurality of words and / or a part of speech of the plurality of words. 8. A function as a means for receiving, as search conditions, information defining the order of appearance of the plurality of words, in addition to the information to be performed.
Computer program as described. 9. The computer according to claim 8, wherein the computer functions as a means for receiving, as a search condition, information defining the number or presence or absence of a non-search target word intermingled between the plurality of words. program. 10. When the list of parts of speech in which the form of a word changes according to the grammar includes a list of the original form of the word and a list of irregular forms of the word, the computer matches the search condition including information related to the part of speech. Means for determining whether or not the word of interest is a regular variation, if the word of interest that is the determination target of whether to perform is not present in any of the original form and the list of irregular variations, Means for inversely generating a character string of the original form from the word of interest when it is determined that the character string is a regular variation; and determining whether or not the character string of the inversely generated original form exists in the list. The computer program according to any one of claims 6 to 9, wherein the computer program is caused to function as means for determining whether a word is an object to be searched. 11. A computer program for causing a computer to search for a document by referring to a list in which words are registered for each part of speech, and to cause a computer to extract an object that matches the search condition, is readable by the computer. In a recording medium, a program code means for causing a computer to delimit a document for each sentence, and a program for causing a computer to receive, as a search condition, part or all of the object and / or information relating to the part of speech of the object Code means, and a computer for searching for a document by referring to the list,
A program code for extracting and presenting a sentence including an object matching the search condition, and a program code for causing the sentence to be presented.