JPH05314180A - Information retrieving device - Google Patents

Abstract

PURPOSE:To automatically know the intention of a user about a retrieval request by automatically deciding the coefficient of a function corresponding to a fuzzy connection operator so as to be matched with the intention of the user based on an evaluation value corresponding to a retrieved result and displaying the degree of consideration of a fuzzy proposition in a retrieved question found from the pertinent coefficient. CONSTITUTION:An adaptation arithmetic part 102 finds the adaptation of data corresponding to the fuzzy proposition in the retrieved question. A variable connection arithmetic part 103 inputs the adaptation, operates a functional arithmetic operation having the coefficient, and finds the total evaluation value corresponding to the entire retrieved questions. A coefficient adjusting part 105 adjusts the coefficient of the functional arithmetic operation of the variable connection arithmetic part 103 so that an error between the total evaluation value searched by the variable connection arithmetic part 103 and the evaluation inputted by the user can be the minimum. The degree of consideration arithmetic part 106 finds the degree of consideration of the fuzzy proposition in the retrieved question from the coefficient of the functional arithmetic operation of the variable connection arithmetic part 103, and the degree of consideration display part 107 displays the degree of consideration.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval device for retrieving predetermined information from a database.

[0002]

2. Description of the Related Art Conventionally, as a method for searching a database using an ambiguous search question sentence, for example, Information
There was a method of using the membership function as shown in Processing & Management, No.13, pp.289-303. FIG. 2 shows a block diagram of an example of a conventional information retrieval device using a membership function. In FIG. 2, 201 is a database that stores data to be searched, 202 is a data retrieval unit that sequentially retrieves data from the database 201, and uses a membership function to obtain a data matching degree for a fuzzy proposition in a search question sentence. , 203 is a min corresponding to the fuzzy combination operator of “Katsu” and “Or” in the search question sentence, with the degree of matching of the data as an input.
A combination operation unit that performs an operation or a max operation to obtain a total evaluation value of data for the entire search question sentence, and 204 is a search result display unit that displays data as search results in descending order of the total evaluation value obtained by the combination operation unit 203. is there.

As an example of data stored in the database 201, an accommodation facility database is shown in (Table 1).

[0004]

[Table 1]

The accommodation facility database 201 has "hotel name", "time required" from the airport, "number of rooms", etc. as attributes. Further, as an example of the membership function stored in the goodness-of-fit calculation unit 202, μ _{TIME = SHORT} representing “required time = short” and μ representing “number of rooms = large”
_{ROOM = MANY} is shown in FIG.

A description will be given below on the assumption that a staff member of a travel agency uses the conventional information retrieval apparatus configured as described above to retrieve an accommodation facility desired by a customer. Now, the customer wants "accommodation that has a short duration from the airport and a large number of rooms", so the staff member "search for accommodations that have a short duration from the airport and a large number of rooms." Suppose that a search is made with a search question sentence composed of a fuzzy proposition and a fuzzy combination operator (in this example, "and" is used).

The degree-of-fit calculation unit 202 is used for the database 2
All accommodation data D _i from 01 (i is the number of data) taken out in the order, by using the membership function of FIG. 3, fuzzy proposition as "required time = short", "Number of rooms = often" and for the data D _{i Goodness of} fit μ _{TIME = S HORT} (D _i ) and μ
_{ROOM = MANY} (D _i ) is obtained and sent to the join operation unit 203. Μ _{TIME = SHORT} (D _i ) and μ _{ROOM = MANY} (D _i ) of each accommodation data
And (Table 2) and (Table 3), respectively.

[0008]

[Table 2]

[0009]

[Table 3]

Next, the join operation unit 203 uses a fuzzy join.
By performing processing on the operator "Katsu", search quality can be improved.
Overall evaluation value of the data Di for the whole question sentence μ (D_i),
μ (D _i) Is sent to the search result display unit 204. Where "
As for the processing for "tsu", as shown in (Equation 1), fuzzy
For the propositions "time required = short" and "number of rooms = large"
Data D_iGoodness of fit μ_{TIME = SHORT}(D_i) And μ
_{ROOM = MANY}(D_i) And are input to perform min operation.
And match the results to the entire search query of data Di
Degree μ (D _i).

[0011]

[Equation 1]

Μ (D _i ) of each accommodation facility data is shown in (Table 4). The search result output unit 204 outputs data as a search result in the descending order of the matching degree μ (D _i ) with respect to the entire search question sentence. The search results are shown in (Table 5).

[0013]

[Table 4]

[0014]

[Table 5]

[0015] In addition, as in "Search for accommodation facilities where the required time from the airport is short or there are many rooms",
When a search is performed with a search question sentence having a different fuzzy join operator compared to the above search example, the join operation unit 203 replaces the min operation, which is the process for "Katsu", with
Using the max operation as the processing for “or” μ
Find (D _i ). The other processing is the same, so detailed description will be omitted.

[0016]

[Problems to be Solved by the Invention] In the above example, the degree of importance of which of the two fuzzy propositions, "the time required from the airport is short" and "there are many rooms", which the customer desires, is emphasized. Generally varies from customer to customer. However, in the conventional information retrieval device with the above configuration, since the min operation and the max operation are performed as the processing for the fuzzy join operator of "and" and "or", respectively, the degree of importance of the fuzzy proposition in the retrieval question sentence is increased. Can not be considered. Therefore, in order to select an accommodation facility that the customer is satisfied with from the search results or to present an alternative plan, a staff member must verbally ask the customer the degree of importance.

The present invention solves the above-mentioned conventional problems, and automatically obtains the degree of importance of each fuzzy proposition in a search question sentence from the evaluation value given to the search result by the customer,
An object of the present invention is to provide an information retrieval device capable of knowing the degree of importance of a customer to a fuzzy proposition by displaying the degree of importance.

[0018]

In order to achieve this object, the present invention uses a database storing data to be searched and a membership function to determine the degree of conformity of the data to the fuzzy proposition in the search query. And a variable coupling operation unit that receives the degree of fitness as an input and obtains a comprehensive evaluation value of the data for the search question sentence by a function operation having a coefficient corresponding to a fuzzy combination operator in the search question sentence. A search result display unit that displays the data and the comprehensive evaluation value as a search result; a coefficient adjusting unit that adjusts a coefficient of the function operation using the comprehensive evaluation value and a given user evaluation value; An arrangement is made up of an importance degree calculation unit that obtains the importance degree of the fuzzy proposition in the search question sentence from the coefficient of the function calculation, and an importance degree display unit that displays the importance degree. It is.

[0019]

With this configuration, the present invention obtains the degree of matching of each data in the database with respect to each fuzzy proposition in the search question sentence when a search question sentence including at least two fuzzy propositions is input. By inputting these matching degrees and performing a functional operation having a coefficient corresponding to the fuzzy combination operator "and" or "or", the comprehensive evaluation value of the data for the entire retrieval question sentence is obtained. An error between the total evaluation value and the user evaluation value input by the user is obtained, and the coefficient of the function operation is adjusted using the error. Since the degree of importance of the fuzzy proposition in the retrieval question sentence is automatically obtained and displayed from the coefficient of this function operation, it is possible to know the idea of the degree of importance of the user.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows a block diagram of an information retrieval apparatus which is an embodiment of the present invention. In FIG. 1, 101 is a database that stores the data to be searched, 102 is the data that is extracted from the database 101 in order, and the matching degree calculation unit that uses the membership function to find the matching degree of the data with respect to the fuzzy proposition in the search question sentence , 103 are input variables of the goodness-of-fit degree output from the goodness-of-fit calculation section 102, and obtain a total evaluation value by a function operation having a coefficient corresponding to the fuzzy combination operator of "and" or "or" in the search question sentence. A combination calculation unit, 104 is a search result display unit that displays data as a search result in the order of high total evaluation value output from the variable combination calculation unit 103, and 105 is a variable degree of the conformity degree output from the conformance degree calculation unit 102. Join operation unit 103
A coefficient adjustment unit that adjusts the coefficient of the function operation of the variable combination operation unit 103 using the total evaluation value that is the output of the variable operation operation and the user evaluation value input by the user. The importance degree calculation unit for obtaining the importance degree of each fuzzy proposition in the search question sentence from the coefficient of
The importance degree display unit displays the importance degree calculated by the importance degree calculation unit 106.

As an example of the data stored in the database 101, the same accommodation facility database (Table 1) as the conventional example is used. As an example of the membership function stored in the goodness-of-fit calculation unit 102, the same membership function of FIG. 3 as in the conventional example is used.

The operation of the information retrieving apparatus of the present embodiment configured as described above is assumed by a staff of a travel agency to retrieve the accommodation facility desired by the customer, as in the conventional example. Explained. Now, the same as in the conventional example, a search question consisting of a fuzzy proposition "Search for accommodations with a short period of time from the airport and a large number of rooms" and a fuzzy combination operator (in this example, "and"). Let's say you search in.

The goodness-of-fit calculation unit 102 obtains the goodness-of-fit μ μ _{= TIME} ( _SHO ) (D _i ) and μ of the data D _i (i is a data number) for the fuzzy propositions “time required = short” and “number of rooms = large”. The processing up to _{obtaining ROOM = MANY} (D _i ) and sending it to the variable combination calculation unit 103 is the same as in the conventional example, so detailed description will be omitted. Μ _{TIME = SHORT} (D _i ) and μ of each accommodation data
_{ROOM = MANY} (D _i ) is the same as the conventional example (Table 2) (Table 3). The variable combination calculation unit 103 inputs the goodness of fit μ _{TIME = SHORT} (D _i ) and μ _{ROOM = MANY} (D _i ),
By performing the operation of (Equation 2) as a process corresponding to the fuzzy join operator “Katsu”, the comprehensive evaluation value μ (D _i ) for the entire search question sentence of the data D _i is obtained, and μ (D _i ) is searched. It is sent to the result display unit 104.

[0024]

[Equation 2]

The search result display unit 104 outputs data as a search result in the descending order of the total evaluation value μ (D _i ) for the entire search question sentence. As an initial value, the coefficient of (Equation 2) is p ₀ = p
Table 6 shows the search results when ₁ = p ₂ = 0.01, p ₃ = 5, and p ₄ = 5 are set.

[0026]

[Table 6]

Here, if the customer determines that the search result in (Table 6) does not match his or her search request, the customer makes a search request for one or a plurality of data D _k (k is a data number). If a user evaluation value μ ^T (D _k ) that is considered to match is input, the coefficient adjustment unit 105 automatically adjusts the coefficient of the function operation of the variable combination operation unit so that the search result matches the user's search request. To do. Now, it is assumed that the customer inputs the user evaluation value of (Table 7) as the comprehensive evaluation that matches his search request for the accommodation facilities E and C.

[0028]

[Table 7]

The coefficient adjusting unit 105 performs the following processes from [Step 1] to [Step 4] based on the steepest descent method, so that the error between the output of the variable combination calculating unit 103 and the user evaluation value is calculated. Adjust the coefficient p _h (h = 0,1,2,3,4) of the function operation so that the sum of squares is minimized.

[Step 1] The variable combination operation unit 103
A comprehensive evaluation value μ (D _k ) for the entire search question sentence of the data D _k for which the user inputs the evaluation value is calculated.

[0031] [Step 2] coefficient adjusting unit 105 (the number 3) and Delta] p _h calculated from the equation (4) and (5), and updates the p _h by adding Delta] p _h the parameter p _h.

[0032]

[Equation 3]

[0033]

[Equation 4]

[0034]

[Equation 5]

[Step 3] The variable combination operation unit 103 and the coefficient adjusting unit 105 are changed from [Step 1] to [Step 2].
The processes up to are performed for each data D _k for which the user has input the user evaluation value.

[Step 4] The variable combination calculation unit 103 and the coefficient adjustment unit 105 are connected to [Step 1] to [Step 3].
The processes up to are performed until the sum of squares of the error represented by (Equation 6) becomes less than or equal to a certain threshold value.

[0037]

[Equation 6]

In this example, as a result of performing the processing from [Step 1] to [Step 4], p ₀ = 0.0, p ₁ = 0.3, p ₂ =
Adjusted to 0.7, p ₃ = 1, p ₄ = 1. Variable combination calculation unit 103
After the adjustment of the coefficient, calculates again the comprehensive evaluation value for the search question sentence. Finally, the search result display unit 104 outputs the data as a search result in descending order of the total evaluation value. The search results using the adjusted coefficients are shown in (Table 8).

[0039]

[Table 8]

Further, the degree-of-importance calculating unit 106 uses each of the fuzzy propositions “required time = short” and “room” by using (Equation 7) and (Equation 8) from the coefficient of the function operation of the variable combination operation unit 103. number = determine the importance degree w _1, w ₂ of the many. "

[0041]

[Equation 7]

[0042]

[Equation 8]

The degree of importance w ₁ and w ₂ respectively emphasizes that "requirement time = short" in the search question sentence "search for accommodation facilities where the required time from the airport is short and the number of rooms is large". This indicates the degree of importance of whether "the number of rooms = many" is emphasized. Degree of importance from (Equation 7) and (Equation 8)
The sum of w ₁ and w ₂ is always 1. The importance degree display unit 107 displays the importance degree calculated by the importance degree calculation unit 106. The display result of the importance degree display unit 107 is shown in (Table 9).

[0044]

[Table 9]

From (Table 9), the staff of the travel agency can know that the customer attaches importance to "the number of rooms = large" at a ratio of 7 to 3.

As described above, according to the present invention, the coefficient adjustment unit for adjusting the coefficient of the variable combination calculation unit using the total evaluation value obtained by the combination calculation unit and the input user evaluation value, and the variable combination unit Automatically knowing the degree of importance of a customer to a fuzzy proposition by providing a degree-of-importance calculation section that obtains the degree of importance of a fuzzy proposition in the search question from the coefficient of the calculation section and an importance degree display section that displays the degree of importance Therefore, the clerk can efficiently perform highly reliable service.

In the embodiment of the invention, the steepest descent method is used as the method of adjusting the coefficient of the function operation, but other methods such as the quasi-Newton method may be used.

[0048]

As described above, according to the present invention, the coefficient adjusting unit for adjusting the coefficient of the variable combination calculating unit using the comprehensive evaluation value obtained by the variable combination calculating unit and the input user evaluation value, and the variable adjusting unit By providing a degree-of-importance calculation section that obtains the degree of importance of the fuzzy proposition in the search question from the coefficient of the join operation section and an importance degree display section that displays the degree of importance, the degree of importance of the customer to the fuzzy proposition is automatically known. Therefore, the staff can efficiently perform highly reliable service.

[Brief description of drawings]

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

FIG. 2 is a block diagram of an example of a conventional information search device.

FIG. 3 is a diagram of a membership function used in a conventional information retrieval device.

[Explanation of Codes] 101,201 Database 102,202 Fitness degree calculation section 103 Variable combination calculation section 203 Combination calculation section 104,204 Search result display section 105 Coefficient adjustment section 106 Importance degree calculation section 107 Importance degree display section

Claims (1)

[Claims] 1. A database for storing data to be searched, a membership function, and a goodness-of-fit calculation unit for obtaining a goodness of fit of the data to a fuzzy proposition in a search question, and the goodness of fit as inputs. A variable combination operation unit that obtains a comprehensive evaluation value of the data for the search question sentence by a functional operation having a coefficient corresponding to a fuzzy combination operator in the search question sentence, and a search that displays the data and the comprehensive evaluation value as a retrieval result. From a result display unit, a coefficient adjustment unit that adjusts the coefficient of the function operation using the overall evaluation value and a given user evaluation value, and the coefficient of the function operation,
An information retrieval apparatus comprising: an importance degree calculation unit for obtaining an importance degree of a fuzzy proposition in a search question sentence; and an importance degree display unit for displaying the importance degree.