JP2001273322A - Information retrieving method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information retrieving method for retrieving information under the consideration of the meaning of qualifier at the time of using a retrieval key including qualifier such as adjective. SOLUTION: This information retrieving method comprises a retrieval key separating process 6 for separating information inputted as a retrieval key into qualifier and a word to be qualified, a relevant information for qualifier outputting process 7 for outputting information expressing the qualifier separated by the retrieval key separating process 6 with another expression, a relevant information for a word to be qualified outputting process 8 for outputting a term for expressing the situation, state, and structure of an object expressed by the word to be qualified separated by the retrieval key separating process 6, a comparing process 9 for comparing the information outputted by the relevant information for qualifier outputting process 7 with the information outputted by the relevant information for a word to be qualified outputting 8, and when any word is matched, judging that the both words are relevant information, and extracting the information, and a retrieval process 10 for retrieving the information by using at least the word to be qualified and the information extracted by the comparing process 9 as a keyword.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information search method in a case where a keyword includes a modifier such as an adjective in a search for information using a keyword.

[0002]

2. Description of the Related Art With the spread of the Internet, information can be extracted not only from a database dedicated to data search but also from various places such as respective homepages.

In a conventional general search device, such a database is searched only for a character string that matches a character string specified by an operator. For example, "large house"
If the character string "wide house" is described in the search target database, the information and the information accompanying it are detected.

[0004] Further, the invention of "Japanese Patent Application Laid-Open No. 10-207896 and method and apparatus for retrieving search terms, and information retrieving method and apparatus" for retrieving using not only a character string specified by an operator but also a character string related to the character string has been developed. It is considered.

FIG. 13 is a flowchart showing the principle of the conventional invention. In the conventional invention, an input search key is subjected to morphological analysis, and a term expanded by converting or adding another word to a synonym or a synonym of a word constituting the search key or deleting a certain character is used. create. For example, when a search key of “Spring / Autumn Publishing Co., Ltd.” is input, “Spring / Autumn Publishing”, “Spring / Autumn Publishing Company”, “Spring / Autumn Publishing Co., Ltd.”, etc. are created as extended terms. An information search apparatus that uses the created term as a new search key is a conventional invention.

[0006]

However, when a search key including a word indicating the state of the search target is used, in the conventional invention, each of the words constituting the search key is compared with the search key input by the operator. Is converted to a synonym or a synonym, or replaced with a word of a higher concept to form a new search key. For example, if an operator wants to search for "a house built on a large land", a search for "a wide house" When a key is entered, a new search key such as "large house" or "wide building" is generated. For the word "wide" indicating the state of the search target, characters indicating a specific size are not It cannot be converted.

As a result, "wide" representing the state of the search object
Necessary information cannot be detected unless the character itself or “big” is used, and even if there is an information source (database) containing quantitative values, that database should be used effectively. Can not.

Therefore, in the case of a character string in which a word to be modified (modifier) and a word to be modified (modified word) such as “wide house” are combined, information containing the character string is simply extracted from the database. Instead, search for information that indicates what it means, taking into account what it means to be a "wide house" (for example, how much space the house has). An apparatus and method for information retrieval as a key are required.

[0009] In the case of a specialized database relating to land information, appropriate information can be extracted by specifically designating the area of the land, but in this case, only a dedicated database can be used and it is scattered around the world. Information cannot be used effectively.

Accordingly, an object of the present invention is to provide an information retrieval method for retrieving a database from a character string composed of a modifier and a modifier by considering the meaning of each word.

[0011]

In order to solve this problem, an information retrieval method according to claim 1 of the present invention provides a method for modifying a modifier entered and a modifier to be modified for information input as a retrieval key. And a search step of searching for information using at least the related information for a modifier associated with the modifier and the qualified word as a search key.

As described above, the search key separating step of separating the information input as the search key into the modifier to be modified and the modifier to be modified, and at least the relation for the modifier associated with the modifier. A search step of searching for information using the information and the qualified word as a search key,
For example, in a case where information is searched for in “wide house”, the search key separating unit separates the character string “wide house” into the modifier “wide” and the qualified word “house”. This separation method is performed by natural language processing using general syntax analysis. As a result, by using a search key including a modifier such as an adjective, information in consideration of the meaning of the search key can be searched from a local or network database.

[0013] The information retrieval method according to the second aspect is the first aspect.
In, the modifier-related information associated with the modifier is a numerical value, a chemical formula, a technical term, or a word of another expression that quantitatively expresses the modifier. In this way, the related information for a modifier associated with a modifier is a numerical value, a chemical formula, a technical term, or a word of another expression that quantitatively expresses the modifier, for example, when the modifier is “wide” , "10m ² -20
m ² ”,“ area ”, and the modifier“ smell ”,“ (C
H ₃ ) ₃ -N ”and when the modifier is“ red ”,“ FF000
Modifier related information such as "0" is associated.

According to a third aspect of the present invention, there is provided an information retrieval method comprising: a retrieval key separating step of separating information input as a search key into a modifier to be modified and a modifier to be modified; And a modifier-related information output step of outputting information expressing the modified modifier in another expression, and, among the information output from the modifier-related information output step, a qualified word separated in the search key separating step. And a search step of searching for information using at least the information extracted in the comparison step and the qualified word as a search key.

As described above, the modifier related information output step for outputting information representing the modifier separated in the search key separating step in another expression, and the information output from the modifier related information output step, And a comparison step of extracting information related to the qualified word separated in the search key separating step and the relevant information for the qualified word. , “Area”, and “10 m ² ”, and when non-quantitative information is used, for example, the modifier “red” corresponds to non-quantitative information such as “color” and “FF0000”. In the comparison step, information related to the word to be modified is extracted from the information. Then, information is retrieved from the database using the character string extracted in the comparison step and the modifiers separated in the search key separation step.

Thus, the character string having an ambiguous expression is converted into a character string having a clear expression, so that necessary information can be more accurately searched.

According to a fourth aspect of the present invention, there is provided an information retrieval method comprising: a search key separating step of separating information input as a search key into a modifier to be modified and a modifier to be modified; And a modifier-related information output step of outputting a numerical value and a term indicating what the numerical value indicates, as the information for quantitatively expressing the qualified modifier, and a qualified word separated in the search key separating step. The situation of the object represented by
A related word output process for outputting a term for expressing a state and a structure, and comparing the information output by the qualified word related information output process with the information output by the qualified word related information output process. If there is a matching word, it is determined that the information is related to each other, and the comparing step of extracting the information and the searching step of searching for information using at least the qualified word and the information extracted in the comparing step as keywords are performed. Including.

As described above, in the retrieval key separating step,
The numerical value and the
For a qualifier that outputs a term indicating what the number indicates
The related information output process and the
For expressing the situation, state, and structure of the object represented by the modifier
Related information output process for qualified words that outputs words, and for qualified words
Information output by the related information output process and related information for the qualified word
Compare the information output by the output process and find a match.
A comparison engine that determines that information is related to each other and extracts the information
In the step of outputting relevant information for modifiers, the modifier
For example, a character string that quantitatively expresses a word is output. Modifier
Is "wide", "10m^Two~ 20m ^Two”And“ plane ”
Quantities such as "product" and what the numbers indicate
Output a plurality of combinations with terms that indicate

In the word-related information output step, for example, as to the word "house" as a word to be qualified, terms "area", "color", "price",
"Height" etc. are output. That is, of the two types of character strings extracted in the modifier related information output step, words of the same type as the term representing what the numerical value indicates are output.

In the comparing step, of the character strings (the numerical values and the terms indicating the numerical values) output from the modifier related information output step, an appropriate character string for modifying the qualified word is determined. In order to do so, the terms output from the modifier related information output step and the qualified word related information output step are compared with each other, and the matching term “area” in the character string output from the modifier related information output step is compared. Is extracted.

Then, information is retrieved from the database using the character string extracted in the comparison step and the modifiers separated in the search key separation step.

Thus, for example, even if a computer is input to search for a character string such as "large house", when searching the database, "area", "10 m ^{2 to} 20 m
² ", will be to search for a string called" house ", it is possible to search for information with a sense that the" area of 10m ² ~20m ² home ".

The information retrieval method according to the fifth aspect is the third aspect of the invention.
Or 4, in the numerical value output from the modifier related information output step, includes a quantitative value conversion step of calculating a value smaller or larger than the numerical value, the range indicated by the value calculated by the quantitative value conversion step This is the range of the search target in the search process.

In the modifier related information output step, when only one numerical value is output as a character string to be quantitatively expressed, if the value is slightly different from the value, the character string may not be searched. is there. Therefore, by calculating a value smaller or larger than the numerical value and incorporating a quantitative value conversion step of expanding the range of the numerical value to be searched using those values, information when it is not always necessary to stick to one quantitative value. Search becomes possible. For example, if the value of 50 is output by the related word means for modifiers, the quantitative value conversion means outputs the values of 30 and 70, and the search means searches for a numerical value within the range of 30 to 70. do it.

[0025] The information retrieval method according to claim 6 is directed to claim 5.
, A value smaller or larger than the numerical value output from the modifier related information output step is calculated using a membership function used in fuzzy logic. in this way,
Since a value smaller or larger than the numerical value output from the modifier related information output step is calculated using the membership function used in fuzzy logic, the numerical range of the search target can be expanded.

[0026] According to a seventh aspect of the present invention, there is provided an information retrieval method, comprising: a retrieval key separating step of separating information inputted as a retrieval key into a modifier to be modified and a modifier to be modified; As information, a modifier-related information output step of outputting information representing a color and a term indicating what the information indicates, and a term for expressing a color as the information related to the qualified word. Outputting the relevant information for a qualified word to be output, and comparing the information output by the relevant information output step for a qualified word with the information output by the relevant output process for a qualified word, The method includes a comparing step of extracting the information by determining that the information is certain, and a searching step of searching for information using at least the word to be modified and the information extracted in the comparing step as keywords.

As described above, as the color information related to the modifier, a modifier-related information output step of outputting information representing a color and a term indicating what the information indicates, Related information output step for outputting information expressing a color and a term indicating what the information indicates as information to be performed, and outputting a term for expressing a color as information related to the word to be modified And a related-word-related information output step, so that a modifier for a color can be handled. In addition, not only a quantitative value (quantity) but also a simple character string can be applied as the information on the modifier. In this case, in the modifier related information output step, for example, when the modifier is “red”, “color”, “FF0000”, etc.
A plurality of combinations of non-quantitative information (character strings) and terms indicating what they represent are output. In the modified word related information output step, a term for expressing the color of the object represented by the qualified word is output, and in the same manner as in claim 4, the modified word related information output step and the qualified word related A similar effect can be obtained by extracting and searching for a character string that matches the information output step.

[0028] In the information search method according to the present invention, a search key separating step of separating information inputted as a search key into a modifier to be modified and a modifier to be modified, and a chemical key associated with the modifier. As the information on the substance, a qualifier related information output step of outputting a term indicating what the chemical formula or the name of the substance indicated by the chemical formula and the name of the chemical formula or the substance indicate, and as information related to the qualified word A correlated word-related information output step of outputting a term for representing information on a chemical substance, and the information output by the qualified word-related information output step and the information output by the qualified word-related information output step. Comparing, if there is a matching word, it is determined that the information is related to each other, and the comparing step of extracting the information, and at least the information extracted in the modified step and the comparing step are used as keywords. And a step of search.

As described above, as information on chemical substances related to modifiers, modifier-related information output for outputting a chemical formula or the name of a substance represented by a chemical formula and a term indicating what the chemical formula or substance name indicates. A process and a related word output process for outputting a term for representing information of the chemical substance as information related to the word to be modified, and a chemical substance information as information related to the word to be modified. And a process of outputting related information for a qualified word for outputting a term for the word, so that a modifier related to an odor can be handled. In this case, in the modifier related information output step, for example, the modifier is “smell”
In the case of, a plurality of combinations of non-quantitative information (character strings) such as “smell” and “ammonia smell” and a term indicating what it indicates are output. In the modified word related information output step, a term for expressing the odor of the object represented by the qualified word is output, and in the same manner as in claim 4, the modified word related information output step and the modified word related A similar effect can be obtained by extracting and searching for a character string that matches the information output step.

[0030] According to a ninth aspect of the present invention, there is provided an information retrieval method, comprising: a search key separating step of separating information input as a search key into modifiers to be modified and modifiers to be modified; A modifier-related information output step of outputting a numerical value and a term indicating what the numerical value indicates as the information to be expressed; a synonym output step of outputting a word similar to the qualified word; A modified word-related information output step of outputting a term for expressing a situation, a state, and a structure of an object represented by the word output from the word output step; and the information output by the qualified word-related information output step, and Comparing the information output by the correlated word related information output step, if there is a matching word is determined to be related information together, the comparison step of extracting the information, at least the word and the word from the similar word output step ratio Search information extracted information in step as a keyword and a search process.

As described above, since a synonym output step of outputting a word similar to the word to be modified is included, when "house" is input as a word to be modified, "house" and "apartment" are similar. However, there is no problem that “house” is used as a search key and information about “apartment” is ignored. That is, by incorporating a similar word output step of outputting a word similar to the qualified word so that information similar to the qualified word separated by the search key separating unit can be searched, a plurality of Can be used as a search key, so users do not have to worry about what words should be used as search keys,
Also, the trouble of inputting many search keys can be reduced.

According to a tenth aspect of the present invention, there is provided an information retrieval method, comprising: a search key separating step of separating information input as a search key into a modifier to be modified and a modifier to be modified; As information, a modifier-related information output step of outputting information representing a color and a term indicating what the information indicates, a synonym output step of outputting a word similar to the qualified word, A word-related information output step for outputting a term for expressing a color as information related to the word output from the similar word output step; and the information output by the word-related information output step and the word output. The information output by the modifier related information output step is compared, and if there is a matching word, it is determined that the information is related to each other, and the comparing step of extracting the information; and at least the word and the ratio from the similar word output step are extracted. Search information extracted information in step as a keyword and a search process.

In this way, the same configuration as in claim 7, which can also handle modifiers related to color, includes a similar word output step of outputting a word similar to the word to be modified.
Similar to the ninth aspect, a plurality of similar qualified words can be used as a search key from one qualified word.

An information retrieval method according to claim 11, wherein a search key separating step of separating information input as a search key into a modifier to be modified and a modifier to be modified, and a chemical key associated with the modifier. As the information on the substance, a qualifier related information output step of outputting a chemical formula or a name of the substance indicated by the chemical formula and a term indicating what the chemical formula or the name of the substance indicates, and the qualifier or the modifier. A similar word output step of outputting a similar word, and a modified word related information output step of outputting a term for representing information of a chemical substance as information related to the word output from the similar word output step, The information output by the modifier related information output step and the information output by the modifier related information output step are compared, and if there is a matching word, it is determined that the words are related information, and the information is determined. A comparison step of extracting, and a search step of searching for information information extracted words and the comparing step at least the similar word output process as a keyword.

As described above, in the structure of claim 8, which can also handle modifiers relating to odors, a similar word output step of outputting words similar to the word to be modified is included. A modifier can use a plurality of similar modifiers as search keys.

[0036]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described with reference to FIGS. FIG. 1 is a functional block diagram of an information search device using the information search method according to the first embodiment of the present invention.

In FIG. 1, reference numeral 1 denotes a device for inputting a character string to be used for search, such as a computer keyboard. Reference numeral 2 denotes a device for outputting a search result, for example, a display used in a computer. Reference numerals 3 and 4 denote databases storing information necessary for the apparatus of this embodiment, and reference numeral 5 denotes a database storing various information. Reference numeral 6 denotes a search key separating unit, and the input device 1
Words to modify for the character string entered from (modifier)
And the word to be modified (modified word). Reference numeral 7 denotes a modifier related information output unit that extracts information related to the modifier separated by the search key separating unit 6 from the database 3. Reference numeral 8 denotes a modified word related information output unit that extracts information related to the qualified word separated by the search key separating unit 6 from the database 4. 9
Is a comparing means, which is a means for detecting information that matches both of the information extracted by the modifier related information output means 7 and the modifier related information output means 8. Reference numeral 10 denotes a search unit, which uses the information and the modifiers selected by the comparison unit 9 to
Information related to the character string input by the input device 1 is extracted from the database 5 and a search result is output to the output device 2.

An important feature of the information retrieval apparatus having the above configuration is that it has a retrieval key separating unit 6, a modifier related information output unit 7, and a qualified word related information output unit 8.

Next, the operation of the information retrieval apparatus thus configured will be described below.

The search key separating means 6 is a means for dividing a character string input by the input device 1 into a modifier and a modifier according to the grammar of a language using morphological analysis. For example,
When a character string “wide house” is input, “wide” is used as a modifier and “house” is used as a modifier.

As described above, by separating the character string into the modifier and the modifier, preparation for converting the modifier into a quantitative expression by the modifier related information output means 7 described later is prepared.

In this embodiment, a general morphological analysis is used to separate a character string. However, if a character string can be separated into a modifier and a modifier, any analysis method can be used. Is also good. For example, if both morphological analysis and parsing are used, a character string (sentence) such as "I want to find a large house in a quiet city" will be replaced by a "quiet city".
It is also possible to search for information using keywords such as and "large house".

In this case, "quiet town" is the modifier "quiet" and the modifier "city", and "wide house" is the modifier "wide".
And the qualified word "house", the search by "AND" is performed for the search for "quiet town" and the search for "wide house", and the search using sentences becomes possible.

In the following, in the description of this embodiment, a case where "wide house" is input to the input device 1 will be described.

The modifier related information output means 7 extracts two types of information from the database 3 as information relating to the modifier “wide” separated by the search key separating means 6.
Here, the database 3 storing these pieces of information will be described. FIG. 2 is an explanatory diagram showing the structure of a database of modifier related information according to this embodiment.

In FIG. 2, f31 is a field for storing a modifier, f32 is a field for storing a term indicating what the field f33 indicates, f33 and f3.
Reference numeral 4 denotes a field for storing a quantitative value. If the values are stored in both the fields f33 and f34, the quantitative value extracted from the database 3 by the modifier related information output means 7 indicates a certain range.
In this case, the first information is a value that quantitatively expresses the state indicated by the word “wide”. For example, "10
m ² "," 10ha "," 10m ³ "," 1m ³ ~10
m ³ ”,“ 10 m ² to 50 m ² ”, etc., and may indicate one value, or may indicate a value in a certain range. The second information is a term indicating what the first information indicates. For example, "area", "volume", and the like.

The modifier related information output means 7 having such a database 3 uses the modifier obtained by the search key separating means 6 as a search key and outputs the field f
Since the information corresponding to the modifier is extracted from 32-32 to f34, an ambiguous expression such as "wide" is converted into a specific quantitative value.

When a value is stored in both the fields f33 and f34, "~", "-" is inserted between the character string of the field f33 and the character string of the field f34.
A new character string in which a character string such as "-" or "kara" is inserted is automatically generated.

As described above, the modifier related information output means 7
In the example, the modifiers “wide” include, for example, “area”, “10 m ²
２０20 m ² ”as a word used for quantitative expression.

The qualified word related information output means 8 extracts one type of information from the database 4 as information related to the qualified word “house” separated by the search key separating means 6. Here, the database 4 storing these pieces of information will be described. FIG. 3 is an explanatory diagram showing the structure of a database of related information for a qualified word according to this embodiment.
This information is a word used to represent the state, structure, or the like of the object or the like indicated by the word to be modified. In FIG. 3, f41
Is a field for storing a qualified word, and f42 is a field for storing a word used to represent the situation, state, structure, or the like of an object or the like indicated by the qualified word. In this case, for example, in the case of the qualified word “house”, “area”, “height”, “daylight hours”, “color”, and the like are used. These words are similar to the terms stored in the field f32 of the database 3.

The qualified word related information output means 8 having such a database 4 uses the qualified word obtained by the search key separating means 6 as a search key, and matches the qualified word from the field f42 of the database 4. Extract information.

As described above, the qualified word related information output means 8 determines which of the words extracted by the qualified word related information output means 7 is appropriate as information for modifying the qualified word, which will be described later. The information for determination by the comparing means 9 is output.

The comparing means 9 compares the term extracted from the field f32 by the modifier related information output means 7 with the term extracted from the field f42 by the qualified term related information output means 8 and detects the same term.

The retrieval means 10 retrieves information from the database 5 using the following three types of character strings. The first is a term detected by the comparison means 9. The second is a character string composed of fields f33 or f34 corresponding to the term. The third is a modifier word separated by the search key separating means 6.

The database 5 is intended for not only a local database but also various databases having character strings as data such as a database via a network. In addition, not only specialized databases but also websites scattered on the Internet are considered as a kind of database.

As described above, according to this embodiment, by providing the retrieval key separating means 6, the modifier related information output means 7, and the modifier related information output means 8, the modifier such as an adjective can be changed. The included string can be used as a search key. The character string can be converted into a quantitative expression, and a character string with an ambiguous expression is converted into a character string with a clear expression, so that necessary information can be more accurately searched.

Since a character string containing a modifier can be used as a search key, a sentence can also be used as a search key. As a result, a use mode in which a search key is input in a conversation with a computer using a voice input device that is easy to input a sentence, instead of inputting the search key from a keyboard, can be provided. For example, if the computer is instructed to “search a large house”, and “search for” is a search execution instruction, “large house” is a search key. in this way,
The variation of the input environment for specifying the search key can be expanded.

A second embodiment of the present invention will be described with reference to FIGS. FIG. 4 is an explanatory diagram showing an example of data stored in a database used in a modifier-related information output step according to the second embodiment of the present invention;
FIG. 5 is an explanatory diagram showing another example of the data stored in the database used in the modifier related information output step in the second embodiment. In the first embodiment, “wide” is used as a modifier as an example, but the same effect can be obtained when a word representing a color is used as a modifier.

In the following second embodiment, a case where information relating to color is used as a modifier will be described. Also, the first
(Quantity) that quantitatively indicates the modifier used in the embodiment.
Instead, the case where non-quantitative information (character string) is used will be described, and the same effect will be obtained with non-quantitative information as in the second embodiment.

The modifier related information output step outputs, as color information related to the modifier, information expressing the color and a term indicating what the information indicates. The correlated word related information output step outputs a term for expressing a color as information related to the correlated word. In this case, as shown in FIG. 4, the fields f33 and f34 of the database 3 used when the color modifier is used include the wavelengths of the colors (for example, 610 to 700 nm for red and 590 to 610 nm for orange). The value is stored. The feature here is that a wavelength having a certain width is defined for one color. Further, for an ambiguous expression such as "reddish", a wavelength having a wider range is defined. In the second embodiment, information of text characters included in a homepage or a database, data of a wavelength observed by a measuring instrument, and the like are compared with the values of the fields f43 and f44.

Next, as shown in FIG. 5, a case will be described in which the database 11 is used in place of the database 3 in the functional block diagram of FIG. 1 and the quantitative values are not used as in the first embodiment. Here, the modifiers used are those relating to color as described above.

First, in the first half of the field f113 in FIG. 5, a code (for example, (1)
Hexadecimal notation): Red is FF0000, orange is FF45
00) is stored. The feature here is that a character string such as a code is used in the field f113 instead of a quantity. One or more codes are associated with one color. This is because human eyes cannot recognize subtle color differences even with different colors (codes) as image data. Also,
Even more ambiguous expressions such as "reddish" are associated with more and more codes. In the second embodiment, by using the data in the field f113, it is possible to search image data as well.

Next, character string data is stored in the latter half of the field f113 in FIG. This data is a word expressing the modifier of the field f111 in a more limited expression. For example, “reddish” in the field f111 is replaced with “red”, “red”,
Corresponds to "orange".

In the second embodiment in which these data are used, information on text characters included in a homepage or a general database (corresponding to the database 5 in FIG. 1) is used for the data in the field f113. Will compare.

As described above, according to this embodiment, modifiers relating to colors can be handled. In addition, as information on modifiers, not only quantitative values (quantities)
A simple string can also be applied. As an example, a code indicating a color was used. However, a method of handling certain information (eg, a color name) in association with information (eg, a color code used in image data, etc.) expressed by another expression is based on a sound. Data (example:
MIDI data) and other colors. For example, in the case of sound data, the name and storage location of an electronic file in which sound data is recorded are stored in the field f113, or sound digital data (binary code or the like) is stored.
What is necessary is just to store itself.

A third embodiment of the present invention will be described with reference to FIG. FIG. 6 is an explanatory view of an example of data stored in a database used in a modifier related information output step in the third embodiment of the present invention. In the first embodiment, “wide” is used as a modifier as an example.
The same effect can be obtained even when a word representing an odor is used as a modifier.

In the following third embodiment, a case will be described in which information relating to a modifier (smell) relating to human olfaction is used. The feature of the present embodiment is that a chemical formula is used as the information on the odor instead of the information (quantity) quantitatively indicating the modifier used in the first embodiment.

The modifier related information output step outputs, as information on the chemical substance related to the modifier, a chemical formula or a name of a substance indicated by the chemical formula and a term indicating what the chemical formula or the substance name indicates. In the correlated word related information output step, a term for representing information on the chemical substance is output as information related to the correlated word. In this case, as shown in FIG. 6, the database 12 used when using the odor-related modifier is replaced with the database 3 used in the first embodiment.
Use instead of The information about the odor stored in the field f123 uses a chemical formula that can be handled as a character string, and the structure of the database in FIG. 6 is the same as the database 11 in the second embodiment.

For example, (CH ₃ ) ₃ —N is the chemical formula of trimethylamine, and this component represents the smell of rotten fish. This chemical formula is associated with the modifier “odor”. As described above, in the third embodiment, the modifier related to human olfaction is the field f123
Is associated with the chemical formula stored in. Note that, instead of the chemical formula, the name of the substance represented by the chemical formula or the name of an article containing the substance as a component may be used.

In the third embodiment using these data, the data of the field f123 is compared with the information of the text characters contained in the homepage or the database as in the second embodiment. become.

As described above, in the third embodiment, the modifier relating to color has been described, and in the third embodiment, the modifier relating to odor has been described. "," Dazzling ", etc., the numerical value indicating the light amount is stored in the database 3,
Even if it is stored in the database as a modifier for emotion, such as "scary", a chemical formula or name related to nitroglycerin, or a numerical value indicating the danger value of the radioactivity level, or "death" as a character string Often, information about various modifiers can be stored in a database and used in this embodiment.

FIGS. 7 to 1 show a fourth embodiment of the present invention.
Description will be made based on 0. FIG. 7 is a functional block diagram according to a fourth embodiment of the present invention, FIG. 8 is a graph of an example of a membership function used in a quantitative value conversion step according to the fourth embodiment, and FIG. 9 is a fourth embodiment. FIG. 10 is an explanatory diagram showing the structure of a database used in the quantitative value conversion step. FIG. 10 is an explanatory diagram showing parameters necessary for creating the membership function in FIG.

First, the configuration of an information search device using the information search method according to the fourth embodiment will be described. In FIG. 7, reference numeral 1 denotes a device for inputting a character string to be used for a search, such as a computer keyboard. Reference numeral 2 denotes a device for outputting a search result, for example, a display used in a computer. Reference numerals 3, 4, and 14 denote databases storing information necessary for the apparatus of this embodiment, and reference numeral 5 denotes a database storing various information. Reference numeral 6 denotes a search key separating unit that separates a character string input from the input device 1 into a word to be modified (modifier) and a word to be modified (modifier). Reference numeral 7 denotes a modifier related information output unit that extracts information related to the modifier separated by the search key separating unit 6 from the database 3.
Numeral 13 is a quantitative value conversion means for calculating a smaller or larger value than the quantitative value output from the modifier related means 7. Information on what value is calculated is stored in the database 14. Reference numeral 8 denotes a word-to-be-modified word output unit that extracts information related to the word-to-be-qualified separated by the search key separating unit 6 from the database 4. 9 is a comparison means,
This is a means for detecting information that matches all of the information extracted by the modifier-related information output means 7, the quantitative value conversion means 13, and the modifier-related information output means 8. Reference numeral 10 denotes search means, which extracts information related to the character string input by the input device 1 from the database 5 using the information selected by the comparison means 9 and the word to be modified, and outputs a search result to the output device 2. .

In the information search device of this embodiment,
The difference from the first embodiment is that a quantitative value conversion unit 13 having a database 14 is provided. The quantitative value conversion means 13 and the database 14 in the information retrieval apparatus thus configured will be described.

In the quantitative value converting means 13, when the modifier related information output means 7 outputs one quantitative value as a character string expressing quantitatively, a value smaller or larger than the quantitative value is calculated and output. I do.

For example, when the modifier related information output means 7 outputs "50 m ² ", the numeral "50"
Then, the values of “30” and “70” are calculated, and “30 m ² ” and “50 m ² ” are output from the quantitative value conversion means 13. For this calculation, a membership function of fuzzy logic is used. For example, by using a membership function as shown in FIG. 8, the range of numerical values to be searched is expanded. In FIG. 8, a modifier associated information output means 7 is set to X0.
Is applied, and the value of X (X1, X2) corresponding to Yh, which is a half value of Y0, is adopted as the value output from the quantitative value conversion means 13. Here,
Although the value of X corresponding to Yh is used, the range of the search target can be narrowed by using the value of X corresponding to 90% of Y0, and the modifier related information output means 7 calculates Even if the value does not match, the search target can be a value very close to that value. If the value of the numerical value is not so important as a search condition, the value of X corresponding to 10% of Y0 may be used.

It should be noted that the quantitative value conversion means 13 does not convert all the quantitative values extracted from the modifier related information output means 7 but uses the terms detected by the comparison means 9 (the search means 1).
(A term used as a search key with 0)).

Next, as shown in FIG. 9, information on what membership function is used in the quantitative value conversion means 13 is stored in the database 14.
The field f141 is a field for storing a word to be a modifier, the field f142 is a field for storing a term indicating what is calculated by the membership function, and the field f143 is a membership function. Or, a field for storing information indicating the type, and a field f144 is a field for storing parameters necessary for considering the shape of the membership function.

A word is stored in the field f141. This word is searched for a word that matches the word stored in the field f41 of the database 4 extracted by the modified word related information output means 8. Used for

The field f142 stores a term indicating what the quantitative value represents. This term is stored in the field f42 of the database 4 extracted by the qualified word related information output means 8. Used to find a match for the term in question.

A field f143 stores a number as information indicating the membership function. Reference numeral 1 denotes a convex function as shown in FIG.

The field f144 includes the field f1
The parameters necessary for forming the membership function indicated by 43 are stored. For example, when the membership function is a convex function as shown in FIG. 8, as parameters, ΔX1, ΔX2, Y0,
And the value of Yh. ΔX1, ΔX2, and Y0 are parameters for forming a convex membership function. Yh is a parameter for calculating the value of X calculated by the quantitative value conversion means 13 from the membership function of FIG.

For example, the retrieval key separating means 6 outputs "house" as a qualified word, the comparing means 9 outputs "area" as a term indicating a quantitative value, and outputs a related information for a qualified word as a quantitative value corresponding to the area. When “50 m ² ” is extracted from the means 7, the field f1 is added to the database 14 of FIG.
41, “house” and the field f142 “area” are searched, and the corresponding information is extracted from the fields f143 and f144 in the first row. The information extracted from the field f143 is a value “1” indicating a convex membership function as shown in FIG. 8, and the information extracted from the field f144 is ΔX1, ΔX2, Y0, Y separated by commas.
h (“40, 40, 1, 0.5”). Then, a membership function (in this case, a convex function) is created based on the detected information, and a value within a certain range corresponding to “50 m ² ” extracted from the modifier related information output means 7 (this In this case, 30, 70) is calculated from the membership function.

The calculated value is detected by the retrieval means 10 by checking whether the numerical value extracted from the database 5 exists within the range of the calculated value (30 m ^{2 to} 70 m ² ). It is possible to check whether there is information to be provided.

[0085] In this way, the area of 50m ² rather than just the house, the area it is possible to search for the 30m ² ~70m ² house.

As described above, the provision of the quantitative value conversion means 13 makes it possible to retrieve information when it is not always necessary to stick to one quantitative value.

A fifth embodiment of the present invention will be described with reference to FIGS. FIG. 11 shows a fifth embodiment of the present invention.
FIG. 12 is an explanatory diagram showing the structure of a database used in a similar word output step according to the fifth embodiment of the present invention.

First, the configuration of an information search device using the information search method according to the fifth embodiment will be described. In FIG. 11, reference numeral 1 denotes a device for inputting a character string used for a search, for example, a keyboard of a computer. Reference numeral 2 denotes a device for outputting a search result, for example, a display used in a computer. Reference numerals 3, 4, and 16 denote databases storing information necessary for the apparatus of this embodiment, and reference numeral 5 denotes a database storing various information. Reference numeral 6 denotes a search key separating unit that separates a character string input from the input device 1 into a word to be modified (modifier) and a word to be modified (modifier). Reference numeral 7 denotes a modifier related information output unit that extracts information related to the modifier separated by the search key separating unit 6 from the database 3.
Reference numeral 15 denotes a similar word output unit, which extracts a word having a meaning similar to the word to be modified from the database 16. Reference numeral 8 denotes a modified word related information output unit that extracts information related to the qualified word separated by the search key separating unit 6 from the database 4. Reference numeral 9 denotes a comparing unit that detects information that matches both of the information extracted by the modifier-related information output unit 7 and the modifier-related information output unit 8. Reference numeral 10 denotes search means, which extracts information related to the character string input by the input device 1 from the database 5 using the information selected by the comparison means 9 and the word to be modified, and outputs a search result to the output device 2. .

The information retrieval apparatus of this embodiment is different from the first embodiment in that a similar word output means 15 having a database 16 is provided. In the information retrieval apparatus thus configured, the synonym output means 1
5 and the database 16 will be described.

The similar word output means 15 is a means for outputting a word having a meaning similar to the word to be modified extracted by the search key separating means 6. The qualified word and words having a similar meaning are stored in the database 16, and the similar word output unit 15 extracts a word corresponding to the qualified word from the database 16. Here, as shown in FIG. 12, a field f161 of the database 16 is a field for storing a qualified word.
Is also a search key. The field f162 is a field storing a word having a meaning similar to the qualified word in the field f161.

In the first embodiment, for another word having a similar or the same meaning as the word to be modified, information relating to the word could not be extracted. Information written in another word can also be extracted.

Note that the fifth embodiment may be applied to the second and third embodiments.

[0093]

According to the information retrieval method according to the first aspect of the present invention, a retrieval key separating step of separating information inputted as a retrieval key into a modifier to be modified and a modifier to be modified; A search step of searching for information using at least the relevant information for a modifier associated with the modifier and the qualified word as a search key. For example, when searching for information in a "wide house", a search key separation In the means, the character string "wide house" is replaced by the modifier "wide" and the modifier "house".
To separate. This separation method is performed by natural language processing using general syntax analysis. As a result, by using a search key including a modifier such as an adjective, information in consideration of the meaning of the search key can be searched from a local or network database.

According to the second aspect, the modifier related information associated with the modifier is a numerical value, a chemical formula, a technical term, or a word of another expression that quantitatively expresses the modifier. for wide ", if" 10 m ² to 20 m ² "and" area ", the modifier is" smell "," (CH _{3) 3} -
If the modifier is “N” and the modifier is “red”, the modifier-related information such as “FF0000” is associated.

According to the information search method of the third aspect of the present invention, a modifier related information output step for outputting information representing the modifier separated in the search key separation step by another expression,
The information output from the modifier-related information output step includes a comparison step of extracting information related to the qualified word separated in the search key separation step, so that the modifier-related information output step includes, for example, In the case of information shown quantitatively, the modifier “wide” corresponds to “area” and “10 m ² ”, and when non-quantitative information is used, for example, the modifier “red”
"Color" and "FF0000" correspond to non-quantitative information. In the comparison step, information related to the word to be modified is extracted from the information. Then, information is retrieved from the database using the character string extracted in the comparison step and the modifiers separated in the search key separation step. As a result, the character string having an ambiguous expression is converted into a character string having a clear expression, so that necessary information can be more accurately searched.

According to the information search method of the fourth aspect of the present invention, as information for quantitatively expressing the modifiers separated in the search key separation step, a numerical value and a term indicating what the numerical value indicates And a qualified word related information output step of outputting a term for expressing the status, state, and structure of an object represented by the qualified word separated in the search key separating step. Comparing the information output by the modifier-related information output step with the information output by the modifier-related information output step, and, if there is a matching word, judges that the information is related to each other, and extracts the information. And so
In the modifier related information output step, a character string that quantitatively expresses the modifier, for example, is output. If modifier is "broad", "10 m ² to 20 m ^2" and "area" and the like, to a plurality outputs a combination of a term for either numerical and numerically expressed quantitatively indicates what. In the correlated word related information output step, for example, regarding the
"Area" is a term used to quantitatively describe the state of a "house"
It outputs “color”, “price”, “height” and the like. That is,
Of the two types of character strings extracted in the modifier related information output step, words of the same type as the term representing what the numerical value indicates are output. In the comparing step, of the character strings (the numerical values and the terms representing the numerical values) output from the modifier related information output step, an appropriate character string for modifying the qualified word is determined. Compares the terms output from the modifier related information output step and the modified word related information output step, and has a matching term “area” in the character string output from the modifier related information output step. Extract a character string. Then, information is retrieved from the database using the character string extracted in the comparison step and the modifiers separated in the search key separation step. Thus, for example, even if you input to the computer to search for the character string "wide house",
When searching the database, "Area", "10m ² ~
20m ² ", will be to search for a string called" house ", it is possible to search for information with a sense that the" area of 10m ² ~20m ² home ".

According to the information retrieval method described in claim 5 of the present invention, a quantitative value conversion step of calculating a value smaller or larger than a certain numerical value and expanding the range of the numerical value to be searched using those values is incorporated. This makes it possible to search for information when it is not always necessary to stick to one quantitative value.
Therefore, when only one numerical value is output as a character string to be quantitatively expressed in the modifier related information output step, there is no problem that the character string is excluded from the search target if the numerical value is slightly different from the value. For example, 50 in the related word means for a modifier
Is output by the quantitative value conversion means.
May be output, and the search means may search for a numerical value within the range of 30 to 70.

In the present invention, a value smaller or larger than the numerical value output from the modifier related information output step is calculated by using a membership function used in fuzzy logic, so that the numerical range of the search target is expanded. Can be.

According to the information retrieval method of the present invention, as the color information related to the modifier, a modifier for outputting information expressing a color and a term indicating what the information indicates is used. A related information output step, a modifier related information output step of outputting information representing a color and a term indicating what the information indicates as information related to the modifier, and information relating to the modifier. And a related word output process for outputting a term for expressing a color, so that a modifier for a color can be handled. In addition, as information on modifiers, not only quantitative values (quantities)
A simple string can also be applied. In this case, in the modifier related information output step, for example, when the modifier is “red”, non-quantitative information (character string) such as “color” or “FF0000” and terms indicating what it indicates And output multiple combinations. In the modified word related information output step, a term for expressing the color of the object represented by the qualified word is output, and in the same manner as in claim 4, the modified word related information output step and the qualified word related A similar effect can be obtained by extracting and searching for a character string that matches the information output step.

According to the information retrieval method of the present invention, as the information on the chemical substance related to the modifier, the name of the chemical formula or the substance indicated by the chemical formula and what the chemical formula or the name of the substance indicates are indicated. A qualified word related information output step of outputting a term to be represented, a qualified word related information output step of outputting a term for representing information on a chemical substance as information related to the qualified word, and a qualified word related information output step. As related information,
Since the method includes a process of outputting related information for a word to be modified, which outputs a term for representing information on a chemical substance, a modifier related to an odor can also be handled. In this case, in the modifier related information output step, for example, when the modifier is “odor”, non-quantitative information (character string) such as “smell” and “ammonia smell” and what it indicates is indicated. Output multiple combinations of terms. In the modified word related information output step, a term for expressing the odor of the object represented by the qualified word is output, and in the same manner as in claim 4, the modified word related information output step and the modified word related A similar effect can be obtained by extracting and searching for a character string that matches the information output step.

According to the information search method of the ninth aspect of the present invention, a synonym output step of outputting a word similar to the word to be modified is included. Although “house” and “apartment” are similar, “house” is used as a search key, and information about “apartment” is not ignored. That is,
By incorporating a similar word output step of outputting a word similar to the qualified word so that information similar to the qualified word separated by the search key separating means can be searched, a plurality of similar words can be obtained from one qualified word. Since the modified word can be used as a search key, the user does not have to think about what word should be used as the search key, and the input of many search keys can be reduced.

According to the information retrieving method according to the tenth aspect of the present invention, a similar word outputting step of outputting a word similar to the word to be modified in the same configuration as in the seventh aspect capable of handling a modifier relating to color. Therefore, a plurality of similar modifiers can be used as a search key from one modifier in the same manner as in the ninth aspect.

According to the information retrieval method described in claim 11 of the present invention, in the constitution of claim 8, which can also handle modifiers relating to odors, a similar word output step of outputting a word similar to the word to be modified is included. Therefore, similar to the ninth aspect, a plurality of similar qualified words can be used as a search key from one qualified word.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of an information search device using an information search method according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram illustrating an example of a structure of a database of modifier related information according to the first embodiment;

FIG. 3 is an explanatory diagram showing the structure of a database of related information for a qualified word according to the first embodiment.

FIG. 4 is an explanatory diagram showing an example of data stored in a database used in a modifier related information output step according to a second embodiment of the present invention.

FIG. 5 is an explanatory diagram illustrating another example of data stored in a database used in a modifier related information output step according to the second embodiment.

FIG. 6 is an explanatory diagram of an example of data stored in a database used in a modifier-related information output step according to a third embodiment of the present invention.

FIG. 7 is a functional block diagram according to a fourth embodiment of the present invention.

FIG. 8 is a graph of an example of a membership function used in a quantitative value conversion step according to the fourth embodiment.

FIG. 9 is an explanatory diagram showing the structure of a database used in a quantitative value conversion step according to the fourth embodiment.

FIG. 10 is an explanatory diagram showing parameters necessary for creating the membership function in FIG. 8;

FIG. 11 is a functional block diagram according to a fifth embodiment of the present invention.

FIG. 12 is an explanatory diagram showing the structure of a database used in a synonym output step according to the fifth embodiment of the present invention.

FIG. 13 is a flowchart showing the principle of a conventional example.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Input device (keyboard etc.) 2 Output device (CRT etc.) 3 Database provided in modifier related information output means 7 4 Database provided in qualified word related information output means 8 5 Database 6 Search key separation means 7 Modifier related information output means 8 Modified relation information output means 9 Comparison means 10 Retrieval means 11 Database provided in modifier word related information output means 7 12 Modifier related information output means 7 Database 13 Quantitative value conversion means 14 Database provided for quantitative value conversion means 15 Similar word output means 16 Database provided for similar word output means f31 to f34 Fields of database 3 f41 to f42 Fields of database 4 Fields f111 to f113 Database 11 Field f121-f 23 field field f161~f162 database 16 fields f141~f144 database 14 of the database 12

Claims (11)

[Claims] 1. For information entered as a search key,
A search key separating step of separating a modifier to be modified and a qualified word to be modified, and a search step of searching for information using at least the related information for a modifier associated with the modifier and the qualified word as a search key; Information retrieval method including. 2. The information search method according to claim 1, wherein the modifier-related information associated with the modifier is a numerical value, a chemical formula, a technical term, or a word of another expression quantitatively representing the modifier. 3. For information entered as a search key,
A search key separation step of separating a modifier to be modified and a qualified word to be modified, and a modifier related information output step of outputting information representing the modifier separated in the search key separation step by another expression, A comparing step of extracting, from the information output from the modifier related information output step, information related to the qualified word separated in the search key separating step, at least the information extracted in the comparing step and the A search step of searching for information using a modifier as a search key. 4. For information entered as a search key,
A search key separation step of separating a modifier to be modified and a modified word to be modified, and a numerical value and what the numerical value indicates as information for quantitatively expressing the modifier separated in the search key separation step A relational information output step for outputting a term representing the term, and a relation for a qualified term for outputting a term for expressing the situation, state, and structure of an object represented by the qualified word separated in the search key separation step. The information output step, the information output by the modifier related information output step and the information output by the qualified word related information output step, and if there is a matching word, it is determined that the related information together, An information search method, comprising: a comparison step of extracting the information; and a search step of searching for information using at least the qualified word and the information extracted in the comparison step as keywords. 5. A quantitative value conversion step of calculating a value smaller or larger than the numerical value output from the modifier related information output step, and a range indicated by the value calculated by the quantitative value conversion step The information search method according to claim 3 or 4, wherein a range of a search target in the search step is set as a range. 6. The information search method according to claim 5, wherein a value smaller or larger than the numerical value output from the modifier related information output step is calculated using a membership function used in fuzzy logic. 7. For information entered as a search key,
A search key separating step of separating a modifier to be modified and a modifier to be modified, and outputting information representing a color and a term indicating what the information indicates as color information related to the modifier. A modifier-related information output step, a modifier-related information output step of outputting a term for expressing a color as information related to the modifier, and a modifier-related information output step. Comparing the information and the information output by the related word output step, and if there is a matching word, determine that the information is related to each other; and extracting the information; and comparing at least the qualified word and the comparison A search step of searching for information using the information extracted in the step as a keyword. 8. For information entered as a search key,
A search key separation step of separating the modifier to be modified and the modifier to be modified, and information on the chemical substance related to the modifier, the chemical formula or the name of the substance indicated by the chemical formula and the name of the chemical formula or the substance A modifier-related information output step of outputting a term indicating whether or not, and a modifier-related information output step of outputting a term for representing information on a chemical substance, as information related to the modifier, The information output by the modifier related information output step and the information output by the modifier related information output step are compared, and if there is a matching word, it is determined that the words are related information, and the information is extracted. An information search method comprising: a comparison step; and a step of searching for information using at least the information extracted in the modification step and the comparison step as keywords. 9. For information entered as a search key,
A search key separating step of separating a modifier into a modifier and a modifier to be modified; and a modification outputting a numerical value and a term indicating what the numerical value represents as information for quantitatively expressing the modifier. A word-related information output step, a synonym output step of outputting a word similar to the qualified word, and terms for expressing the situation, state, and structure of an object represented by the word output from the synonym output step And the information output by the modifier related information output step and the information output by the modifier related information output step, and if there is a matching word, An information retrieval method including a comparison step of extracting the information by determining that the information is relevant, and a retrieval step of retrieving information using at least the word extracted from the similar word output step and the information extracted in the comparison step as keywords . 10. A search key separating step of separating information inputted as a search key into a modifier to be modified and a modifier to be modified, and expressing a color as information on a color related to the modifier. A related word output step of outputting information and a term indicating what the information indicates, a synonym output step of outputting a word similar to the qualified word, and a synonym output step of outputting from the synonym output step. As the information related to the word, a qualified word related information output step of outputting a term for expressing a color, the information output by the qualified word related information output step, and the qualified word related information output step. The output information is compared, and if there is a matching word, it is determined that the information is related to each other, and the comparing step of extracting the information, and at least the keyword extracted from the similar word output step and the information extracted in the comparing step are keyword. A search step of searching for information as a password. 11. A search key separating step of separating information inputted as a search key into a modifier to be modified and a modifier to be modified, and a chemical formula or a chemical formula as information of a chemical substance related to the modifier. A related information output step for outputting a name of the substance indicated and a term indicating what the chemical formula or the name of the substance indicates, and a synonym for outputting a word similar to the modified word or the modified word An output step, a qualified word related information output step of outputting a term for representing information on a chemical substance as information related to the word output from the similar word output step, and a modifier related information output step Comparing the output information and the information output by the qualified word related information output step, if there is a matching word, it is determined that the information is related to each other, a comparison step of extracting the information, at least A search step of searching for information using the word extracted from the similar word output step and the information extracted in the comparison step as keywords.