JPH0934900A - Information retrieval method and device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the information retrieval method and device in which a characteristic of a retrieval object is detected by designating an object similar to a characteristics of the retrieval object or an object different therefrom and a retrieval person detects a characteristic of a intended object, based on a fuzzy characteristic. SOLUTION: First of all an object group displayed as a sample is used to perform input at an input section 1 to designate an object similar to and different from a retrieval object. Then a characteristic detection section 2 detects which attribute is noticed and how the characteristic of the attribute is designated, based on the difference from the designated object characteristic. Then an object extract section 4 retrieves and extracts an object corresponding to the characteristic detected by the characteristic detection section 2 by retrieving the attribute storage section 3. A characteristic of a fuzzy retrieval object is detected clearly, based on the characteristic of the designated sample object and an object is specified from the detected characteristic to overcome the difference from the category at the retrieval and the difficulty in the characteristic expression in this way.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention According to the present invention, it is difficult to accurately express the characteristics of an object to be searched, and there are objects having characteristics partially similar to or different from the object to be searched. In this case, the present invention relates to an information search method and apparatus suitable for the case. In particular, it is effective for a system related to design selection while looking at samples such as "This part is not like this, it is like this".

[0002]

2. Description of the Related Art Conventionally, as a method for searching information by category, a method of hierarchically selecting and searching a corresponding category according to a category classification prepared in advance, or inputting a category as a keyword Many search methods have been adopted.

[0003]

However, in the above-mentioned conventional information retrieval method of selecting from the category classification prepared in advance, when the category classification prepared in advance does not match the category classification of the searcher, the category selection is performed. Will be difficult. If the category classification that the searcher wants to specify is not prepared, the searcher's intention cannot be reflected well in the search. Furthermore, if the searcher has not been able to clearly classify the characteristics of the target of the search as a category, there is a problem that it is difficult to input the search condition.

The present invention has been made to solve the above problems, and provides an information search method and apparatus for detecting a characteristic of a target intended by a searcher from an ambiguous feature and performing a search. To aim.

[0005]

In order to solve the above problems, an information search method of the present invention is an information search method for searching a target having a specific characteristic from a search target group in which each target has a plurality of attributes. First, an object group that includes the respective characteristics of a plurality of attributes to be searched as a whole is presented as a sample, and next, an object group that does not correspond to one or more characteristics from the object group presented as the sample Input the designation of the corresponding target, then detect which attribute is noted from the difference in the characteristics of the input target and how the property of the attribute is designated, and then detect the attribute. It is characterized in that the objects corresponding to the characteristics are extracted from the search object group.

[0006] Similarly, the information retrieval apparatus of the present invention is an information retrieval apparatus for retrieving an object having a specific characteristic from a retrieval object group in which each object has a plurality of attributes, and accumulates the characteristics of a plurality of attributes for each retrieval object. The storage means to be stored, the input means for designating the target and the target which are not applicable from the target group of the sample with respect to one or more characteristics, and which attribute is determined from the difference in the characteristics of the target specified by the input means. It is characterized by comprising a detecting means for detecting how the characteristic of the attribute is designated, and an extracting means for extracting an object corresponding to the detected characteristic from the accumulating means.

In the above information retrieval apparatus, the detecting means detects the specified characteristic by using the characteristic of a plurality of attributes as a variable, or the linear format in which the variables are weighted, and the discrimination threshold. The extracting means is configured to substitute the value of the characteristic of the object accumulated in the accumulating means into the linear variable and extract the corresponding object using the threshold of the discrimination as the search criterion. It is suitable for facilitating the extraction of the target when there are a large number.

Further, in the above information retrieval apparatus, the detecting means detects the specified characteristic as a plurality of retrieval criteria, and the extracting means extracts the objects corresponding to each of the plurality of retrieval criteria. , An extraction result corresponding to a logical product of the extraction results, an extraction result corresponding to the most search criteria, or an extraction result having a high degree of correspondence weighted according to the priority of the search criteria. Is preferable for more clearly detecting the ambiguous target feature of the target.

In the above-described information retrieval method and apparatus of the present invention, the object group of the sample is used to specify the object similar to the object to be searched and the object different from the object to be searched,
The characteristic of the ambiguous target object to be searched is clarified and detected from the difference in the characteristic of the specified object, and the object corresponding to the detected characteristic is extracted to specify the target object. This overcomes the difference in category classification in search and the difficulty in expressing characteristics.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be specifically described below with reference to the drawings.

<< First Embodiment >> FIG. 1 is a block diagram showing the arrangement of the essential parts of an information retrieval apparatus according to the first embodiment of the present invention. In FIG. 1, 1 is a characteristic input unit, 2 is a characteristic detection unit, 3 is an attribute storage unit, and 4 is a target extraction unit.

The characteristic input unit 1 determines an object selected by the user from the sample object group and whether the selection is similar selection or difference selection, and sends it to the characteristic detection unit 2. The characteristic detection unit 2 reads the attribute data of the relevant target from the attribute storage unit 3, collects the attribute data by the similar target / different target, detects a search criterion for detecting the detection target, and transmits the search criterion to the target extraction unit 4. The attribute storage unit 3 stores a search target group and the attributes of each target. The target extraction unit 4 extracts the corresponding target from the attribute storage unit 3 using the search criteria received from the characteristic detection unit 2.

FIG. 2 is an example of a screen on which the target is displayed as an option, that is, a sample in the characteristic input unit 1.
The user designates a target similar to the target target and a target different from these targets A to F.

FIG. 3 shows an example of the attributes of each target stored in the attribute storage unit 3. In this example, 12 objects A
~ L has three kinds of attributes of shape, color and size, shape attributes have four characteristics of △ ○ □ ◇, and color attributes have two kinds of characteristics of black and white. 3 of large, medium and small in the attribute of Sa
There are different characteristics.

FIG. 4 shows an example of similar / difference target attributes used in the search reference detection by the characteristic detection unit 2. The ◯ of the category indicates the characteristics of the attribute of the object designated as the similar object,
X indicates the characteristic of the attribute of the target designated as the difference target.

An operation example of the above information search apparatus will be described below as an example of the information search method according to the present embodiment. This operation example is performed under the following settings.

(1) In the characteristic input unit 1, it is assumed that a target similar to or opposite to the feature to be searched is selected from the sample target group displayed on the screen.
Objects with similar characteristics shall be described with a ◯ after the target symbol, and objects with different characteristics shall be described with a x after the target symbol. In the present embodiment example, FIG.
It is assumed that the information selected from the search target group is displayed as a selection target, and that B and E have different characteristics and C has been selected as a target having similar characteristics. .

(2) The characteristic detection unit 2 reads all the attributes of the corresponding target from the attribute storage unit 3 and creates a table of the form as shown in FIG. 4 for each similar / different target to determine whether the target is similar or different. A discriminant function for discriminating is calculated as a search criterion. For example, the characteristic detection unit 2 uses the characteristic of each attribute as a variable and calculates the threshold value of the variable. In this embodiment, “size = small” is calculated as the search criterion.

(3) It is assumed that the attribute storage unit 3 stores all search target attributes in a table. In the present embodiment, 12 items A to L as shown in FIG. 3 are stored as search targets, and the attributes of each target are
It is assumed that three characteristics of color and size are stored.

(4) The target extraction unit 4 checks the attribute of the search target stored in the attribute storage unit 3 based on the search criteria received from the characteristic detection unit 2 and extracts the corresponding target as the search result. I shall.

The operation example of this embodiment is as follows.

First, in the characteristic input unit 1, the target groups A to F of FIG. 2 are displayed as samples and presented to the user. Subsequently, when the user selects objects B and E as objects having characteristics different from the object of interest and object C is selected as objects having characteristics similar to the object of interest, they are The information “B ×, E ×, C ◯” is transmitted to the characteristic detecting unit 2. Next, the characteristic detection unit 2 receives the information “B ×, E ×, C ○” and extracts the attributes of the objects B, E, and C from the table of FIG. 3 stored in the attribute storage unit 3. , An attribute table for each similar / difference target is created as shown in FIG. Further, it is detected that the user pays attention to the shape attribute and specifies the characteristic that the attribute is small from the difference in the characteristics of the attribute table for each similar / different object. Specifically, a process for calculating a discriminant function for discriminating similar objects is performed as a search criterion, a discriminant expression “size = small” is calculated as a search criterion, and the discriminant is transmitted to the target extraction unit 4. Next, the target extraction unit 4 uses "size =
"Small" is received, and as a target corresponding to the search criterion "size = small" from the table of FIG.
"C, G, I, K" are extracted.

In the above description of the embodiment, the characteristic input unit 1 describes the object symbol having a similar characteristic by a circle, and the object having a different characteristic by a symbol after the object symbol. However, the description method depends on the notation of the attributes stored in the attribute storage unit 3. In the present embodiment, it is assumed that the information as shown in FIG. 2 is displayed as a selection target, that B and E have different characteristics, and C is selected as a target having similar characteristics. However, the options, the expression method thereof, and the selected information are arbitrary, and a sample different from the search target may be used as the options. When the search target is a sample, the properties of each attribute are selected to be included in any of the targets to be displayed, and it is not always necessary to select the number of combinations of properties. When the number of samples is large, the features, representatives, etc. may be presented and displayed hierarchically.

Although the characteristic detecting section 2 creates a table having a form as shown in FIG. 4 for each object of similarity / difference, the format of the table is arbitrary. The attribute data need not be a table as long as it is similar / difference attribute data. Further, the discriminant function for discriminating between the similar target and the different target is calculated as the search criterion, but the calculation method of the search criterion is arbitrary, and the characteristic of a plurality of attributes is used as the variable, or the variable is set as the priority. The linear format weighted in accordance with the above and the discrimination threshold value may be calculated.

Although the attributes of all the retrieval targets are stored as a table in the attribute storage unit 3, the types and the number of the accumulated targets, the types of the attributes, and the representation method thereof are arbitrary. . When the target of the selection candidate displayed in the characteristic input unit 1 is different from the search target, the table of the search target and the table of the selection candidate are created separately, and the characteristic detection unit 2 generates the table of the selection candidate. The target extraction unit 4 uses a table to be searched.

The target extraction unit 4 checks the attributes of the search target stored in the attribute storage unit 3 based on the search criteria received from the characteristic detection unit 2 and extracts the corresponding target as the search result. However, there may be multiple search criteria. When there are multiple search criteria, objects that satisfy the logical product of the search criteria, objects that meet the most search criteria, or degrees that meet the search criteria are weighted and calculated according to the priority of the search criteria, A target with a large degree is extracted. When the line type and the threshold for discrimination are received as the search criteria from the characteristic detection unit 2, the target extraction unit 4
Substitutes the value of the characteristic of the target accumulated in the attribute storage unit 3 into a linear variable, and extracts the corresponding target based on the discrimination threshold.

As described above, in the information search apparatus and method according to the present embodiment, the search target object is input from the input of the target object having characteristics partially similar to or different from the target object to be searched. The feature is that the characteristic of is clearly detected and the target corresponding to the characteristic is searched. The user can specify characteristics that do not match existing categories (attributes) or characteristics that cannot be accurately expressed in words as search conditions, without converting ambiguous characteristics into words that can be interpreted by a computer. ,
You can search.

<< Second Embodiment >> Next, a second embodiment of the present invention will be specifically described with reference to the drawings.

FIG. 5 is a block diagram showing the structure of the essential parts of an information retrieval apparatus according to the second embodiment of the present invention. In FIG. 5, 10 is an input unit, 11 is an input control unit, 12 is an attribute extraction unit, 13 is an attribute storage unit, 14 is a search reference calculation unit, 15 is a search reference storage unit, 16 is a search extraction unit, and 17 is an output. The control unit, 18 is an output unit. Of each component,
The input unit 10 and the input control unit 11 correspond to the characteristic input unit 1 of FIG.
The attribute extraction unit 12 and the search criterion calculation unit 14 are in the characteristic detection unit 2 of FIG. 1, the attribute accumulation unit 13 is in the attribute storage unit 3 of FIG. 1, the search criterion accumulation unit 15, the search extraction unit 16, the output control unit 17, The output unit 18 corresponds to the target extraction unit 4 of FIG.

The input unit 10 detects an input operation by the user and sends an input signal to the input control unit 11. Input control unit 11
Receives an input signal, determines an object selected by input and whether the selection is a similar selection or a different selection, and temporarily stores the input information. When the input end signal is received as the input signal, the input information accumulated until then is transmitted to the attribute extraction unit 12. The attribute extraction unit 12 reads the attribute information of each applicable target from the attribute storage unit 13, collects the attributes of those targets by similar / different target, and transmits them to the search criterion calculation unit 14. The search criterion calculation unit 14 calculates a criterion for searching the search target from the received similar / difference target data, and stores the reference in the search criterion storage unit 15. The search extraction unit 16 searches for the corresponding target from the attribute storage unit 13 using the plurality of search criteria stored in the search reference storage unit 15, and integrates the search results to calculate the final search result. , To the output control unit 17. Output control unit 17
Performs an information processing process for outputting the target information received as the search result, and outputs it to the output unit 18. FIG.
Is an example of a screen displaying the search result output to the output unit 18.

Below, an example of the operation of the above information retrieval apparatus will be explained as an example of the information retrieval method of the present embodiment. This operation example is performed under the following settings.

(1) In the input section 10, it is assumed that an object displayed as a sample on the screen is clicked with a mouse to select an object whose one characteristic is similar / different from the characteristic to be searched. Single-click to select an object with a characteristic different from the desired characteristic, and double-click to select an object with a characteristic similar to the desired characteristic. Further, the selection of the target when paying attention to the same characteristic can be continuously performed, and the target selection for the same characteristic is finished by pressing the return key from the keyboard. In the present embodiment, it is assumed that the information as shown in FIG. 2 is displayed as a sample selection target, B and E are single-clicked, C is double-clicked, and then the return key is pressed. I shall.

(2) The input control unit 11 recognizes the type of each target selected by the input unit 10 and the number of clicks, represents the type of the target with an alphabetic symbol, x for a single click and ◯ for a double click. Shall be represented by.

(3) The attribute extraction unit 12 reads all the attributes of the relevant target from the attribute storage unit 13 and creates a table as shown in FIG. 4 for each similar / different target.

(4) It is assumed that the attribute storage unit 13 stores the attributes of all search targets including the target displayed by the input unit 10 as a table. In this embodiment, 12 items A to L as shown in FIG. 3 are stored as search targets, and three features of shape, color, and size are stored as attributes of each target. And

(5) It is assumed that the search criterion calculation unit 14 calculates, as a search criterion, a discrimination function for discriminating between a similar target and a different target from the attributes of the similar / different target transmitted from the attribute extraction unit 12. For example, the characteristic of each attribute is used as a variable, and the threshold value of the variable is calculated. In this embodiment, “size = small” is calculated as the search criterion.

(6) It is assumed that the search criteria accumulating section 15 sequentially stores the search criteria transmitted from the search criteria calculating section 14, and sequentially transmits all the search criteria from the search extracting section 16 in response to the access. In this embodiment, “color = black” has already been set.
It is assumed that the search criteria is stored.

(7) The search extraction unit 16 checks the attribute of the search object stored in the attribute storage unit 13 based on each search standard received from the search standard storage unit 15 and sets the corresponding target as the search result. Extract. Furthermore, the logical product of the search results searched based on each search criterion shall be the final search result. Then, the target extracted as the final search result and its attribute are transmitted to the output control unit 17.

(8) In the output control unit 17, the search / extraction unit 1
It is assumed that the search result received from No. 6 is converted into an image based on the attribute and transmitted to the output unit 18.

(9) The output unit 18 displays the image received from the output control unit 17 on the display.
In this embodiment, a screen as shown in FIG. 6 is displayed.

The operation example of the present embodiment is as follows.

First, in the input section 10, the target groups A to F in FIG. 2 are displayed as samples and presented to the user. Subsequently, when the user inputs an input signal for selecting the target B as a target having a characteristic different from the target target by a single click from the input unit 10, the input signal is transmitted to the input control unit 11 as an input signal. . The input control unit 11 converts the input signal, recognizes that the target is B, and selects by single click,
It is temporarily stored as input information "x". Next, when an input signal for selecting the target E as a target having a characteristic different from the target target by a single click is input from the input unit 10, the input signal is input as an input signal.
1 is transmitted. The input control unit 11 converts the input signal, recognizes that the target is E and selection by a single click, and temporarily records it as input information “E ×”. Then, from the input unit 10 to the target C
When an input signal that is selected by double-clicking as a target having characteristics similar to the target target is input, the input signal is transmitted to the input control unit 11 as an input signal. Input control unit 1
1 converts the input signal and recognizes that the target is C and selection by double-clicking, and "C ○"
Is temporarily stored as the input information. And
When the return key is pressed from the input section 10, it is transmitted to the input control section 11 as an input signal. Input control unit 1
1 recognizes that the input for the same characteristic is completed by converting the input signal, and temporarily stores "B" until then.
It is transmitted to the attribute extraction unit 12 as information “×, E ×, C ◯”.

Next, the attribute extraction unit 12 displays "B ×, E ×,
The information "C ○" is received, the attributes of the objects B, E, and C are extracted from the table of FIG. 3 stored in the attribute storage unit 13, and the attribute table for each similar / difference object as shown in FIG. 4 is created. To do. The search criterion calculation unit 14 performs a process of calculating a discriminant function for discriminating a similar target as a search criterion from the table of FIG. 4 received from the attribute extraction unit 12, and determines a discriminant expression “size = small” as a search criterion. It is calculated and transmitted to the search criterion storage unit 15.
The search criterion accumulating unit 15 accumulates the newly transmitted search criterion “size = small” in addition to “color = black” to the already accumulated search criterion.

Next, the search / extractor 16 first sets “color = black” as the search criteria stored in the search criteria storage 15.
From the table of FIG. 3 stored in the attribute storage unit 13 as a target corresponding to the search criterion “color = black”, “A,
"E, F, G, I, J" are extracted. Next, “size = small” is received as a search criterion, and from the table of FIG. 3 accumulated in the attribute accumulating unit 13, “C, G, I, K ”are extracted. And the logical product of the search results of both search criteria is calculated,
“G, I” is extracted as the final search result. Furthermore, this result is combined with each attribute "G: ○ / black / small""I: ◇
"/ Black / small" to the output control unit 17. The output control unit 17 receives the search result and the attribute thereof, creates an image as shown in FIG. 6 based on the attribute, transmits the image to the output unit 18, and outputs the image on the screen by the output unit 18.

In the description of this embodiment, in the input section 10, the object displayed on the screen is clicked with the mouse, but the input means may be a mouse, a keyboard or a touch panel. When selecting the target of the characteristic that is different from the target characteristic, single-clip is used to select the target of the characteristic similar to the target characteristic, but it is selected by double-clicking, but the distinction between similar target / different target is arbitrary. In the case of a two-button mouse, the left and right buttons may be used for distinction, or the mode may be switched using keys or soft buttons. Further, when the same characteristic is focused on, the selection of the target can be continuously performed, and the target selection for the same property is ended by pressing the return key from the keyboard, but the target selection for the same property is ended. The signal input method is arbitrary, and the soft button may be selected. In the present embodiment, the information as shown in FIG. 2 is displayed as the selection target, but the selection information and its expression are arbitrary, and a sample different from the search target may be used. The selected information and order are also arbitrary.

The input control unit 11 recognizes the type of each target selected by the input unit 10 and the number of clicks, but the recognition target depends on the input method of the input unit 10. The expression of the target also depends on the classification notation of the target stored in the attribute storage unit 13.

Although the attribute extraction unit 12 creates a table having a form as shown in FIG. 4 for each of similar / different objects, the format of the table is arbitrary. Further, the attribute data need not be a table as long as it is similar / different attribute data.

In the attribute storage unit 13, the attributes of all search targets including the target displayed in the input unit 10 are stored as a table, but the type and number of the stored targets,
The type of attribute and its expression method are arbitrary. In addition, when the target of the selection candidates displayed in the input unit 10 is different from the search target, the table of the search targets and the table of the selection candidates are created separately, and the attribute extraction unit 12 creates the selection candidate tables. The search and extraction unit 16 uses a table to be searched.

In the search criterion calculation unit 14, the attribute extraction unit 12
A discriminant function for discriminating between a similar target and a different target based on the attributes of the similar / different target transmitted from is calculated as a search criterion, and in the present embodiment, a search criterion “size = small” is calculated. However, the method for calculating the search criteria may be arbitrary, and the characteristic of a plurality of attributes may be used as a variable, or the linear format in which the variables are weighted and the determination threshold may be calculated.

In the search criterion storage unit 15, "color = black" has already been entered.
It is assumed that the search criteria are accumulated, but the accumulated search criteria and the number thereof are arbitrary.

In the search and extraction unit 16, the search reference storage unit 15
The attribute of the search target stored in the attribute storage unit 13 is checked based on the respective search criteria received from, and the logical product of the searched search results is used as the final search result.
The search method is arbitrary, among the search results of multiple search criteria,
The top several objects that meet many criteria or the objects that meet more than a certain number of criteria may be the final search results. Alternatively, a method may be considered in which the priority is specified in advance for each search criterion, the extraction result of each search criterion is weighted based on the priority, and the final search result is obtained from the degree corresponding to the search criterion. Further, although the target and the attribute thereof extracted as the final search result are transmitted to the output control unit 17, a simple method in which the attribute is not transmitted may be considered. When the linear format and the discrimination threshold are received from the search criteria storage unit 15 as one of the search criteria, the search extraction unit 16 stores the value of the target characteristic stored in the attribute storage unit 13 in the variable of the linear format. Substituting
The corresponding target is extracted based on the discrimination threshold.

The output control unit 17 converts the retrieval result received from the retrieval and extraction unit 16 into an image based on the attribute and transmits the image to the output unit 18. However, if the attribute is not transmitted, the name of the target is transmitted. And so on.

Although the output unit 18 displays the image received from the output control unit 17 on the display, it may be transmitted to another terminal via a communication line instead of the display or printed from a printer. One of the search criteria is
As described above, in the information search method and apparatus of the present embodiment, the characteristics of the search target object are input from the input of a plurality of targets having characteristics partially similar to or different from the search target object. The feature is that it is detected clearly and the object corresponding to the characteristic is searched. The user can specify the characteristics that do not match the existing categories or the characteristics that cannot be accurately expressed by words as the search condition, and can search the ambiguous characteristics without converting them to computer-readable words. You can

The present invention has been specifically described above based on the embodiments, but the present invention is not limited to the above embodiments, and various modifications can be made without departing from the spirit of the present invention. Needless to say.

[0055]

As described above, according to the information retrieval method and apparatus of the present invention, the objects which have some of the characteristics of the retrieval purpose among the objects displayed in the sample and the objects which do not have them. By specifying as a set, the characteristic of interest is clearly detected from the difference in those attributes, and the search condition is clearly grasped, so just by selecting a similar target and a different target, It is possible to search for a target object from ambiguous features. Therefore, the user can specify a characteristic that is difficult to be accurately expressed by words as a search condition without using words. Furthermore, since the conditions for specifying and searching while using the characteristics of the sample are similar to the selection behavior that refers to the daily sample, it is easy for humans to construct a search policy and contribute to the improvement of the usability of the system. it can.

In the above information retrieving apparatus, the detecting means detects the specified characteristic by the linear format and the discrimination threshold value, and the extracting means corresponds to the linear format value and the discrimination threshold value for each object. When the target to be determined is extracted and extracted, the target can be easily extracted when there are many attributes.

Further, the detecting means detects the designated characteristic as a plurality of search criteria, and the extracting means calculates the logical product of the extraction results of the corresponding objects, or the extraction results corresponding to the most search criteria, Or, if the extraction result with a high degree of applicability weighted according to the priority of the search criteria is output, the ambiguous target feature can be detected more clearly, and the target target can be more easily identified. You can search for.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a main part of an information search device according to a first exemplary embodiment of the present invention.

FIG. 2 is a diagram showing an example of a screen on which targets are displayed as options in a characteristic input unit in the information search device according to the first embodiment.

FIG. 3 is a diagram showing an example of an attribute of each target stored in an attribute storage unit in the information search device according to the first embodiment.

FIG. 4 is a view showing similarity / similarity used for search reference detection in a characteristic detection unit in the information search device according to the first embodiment.
It is a figure which shows an example of the attribute of a difference object.

FIG. 5 is a block diagram showing a configuration of a main part of an information search device according to a second exemplary embodiment of the present invention.

FIG. 6 is a diagram showing an example of a screen displaying search results output to an output unit in the information search device according to the second embodiment.

[Explanation of symbols]

 1 ... Character input unit 2 ... Characteristic detection unit 3 ... Attribute storage unit 4 ... Target extraction unit 10 ... Input unit 11 ... Input control unit 12 ... Attribute extraction unit 13 ... Attribute storage unit 14 ... Search criteria calculation unit 15 ... Search criteria storage Part 16 ... Search extraction part 17 ... Output control part 18 ... Output part

Claims (4)

[Claims] 1. An information retrieval method for retrieving an object having a specific characteristic from a retrieval object group in which each object has a plurality of attributes. First, an object group having the respective characteristics of a plurality of attributes of the retrieval object as a whole is provided. Present as a sample, then enter the designation of the target that does not correspond to one or more characteristics from the group of objects presented as the sample, and then specify the characteristics of the input object An information search characterized in that which attribute is focused on from the difference and how the characteristic of the attribute is specified is detected, and then a target corresponding to the detected characteristic is extracted from the search target group. Method. 2. An information retrieval apparatus for retrieving an object having a specific characteristic from a retrieval object group in which each object has a plurality of attributes, and accumulating means for accumulating the characteristics of a plurality of attributes for each retrieval object. Input means for designating a non-applicable target and a relevant target from the sample target group for one or more characteristics, and which attribute is noted from the difference in the characteristics of the target specified by the input means, and the characteristic of the attribute An information retrieving apparatus, comprising: a detection unit that detects how to specify an item, and an extraction unit that extracts an object corresponding to the detected characteristic from the storage unit. 3. The detecting means detects the specified characteristic by a linear form in which the characteristics of a plurality of attributes are used as variables, or the variables are weighted, and a discrimination threshold value, and the extracting means is characterized in that 3. The information retrieval device according to claim 2, wherein the value of the characteristic of the object accumulated in the accumulating means is substituted into a linear variable, and the applicable object is extracted using the threshold value of the discrimination as a retrieval criterion. 4. The detecting means detects the specified characteristic as a plurality of search criteria, and the extracting means extracts a target corresponding to each of the plurality of search criteria and obtains a logical product of the extraction results. 4. The corresponding extraction result, the extraction result corresponding to the largest number of search criteria, or the extraction result having a high corresponding degree weighted according to the priority of the search criteria is output. Information retrieval device described.