JPH04281562A - Retrieving device by fuzzy inference - Google Patents

Abstract

PURPOSE:To accelerate processing speed by narrowing down a retrieval target in first retrieval based on a significant condition, and finding the final retrieval result in second retrieval. CONSTITUTION:First retrieval parts 4, 5 which select first retrieval data by retrieving retrieval target data find adaptation for the first retrieval data by fuzzy inference by using a condition selected in advance as a required retrieval condition out of inputted retrieval conditions, and select data with adaptation over a threshold value as the first retrieval data. Second retrieval parts 7, 8 performs prescribed retrieval by the fuzzy inference setting the first retrieval data as a target by using the condition other than the required retrieval condition out of the inputted retrieval conditions. In other words, the retrieval target data is narrowed down based on whether or not the data satisfies the required retrieval condition over constant reference at the first retrieval parts 4, 5, and the retrieval is performed by using another retrieval condition at the second retrieval parts 7, 8.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for searching suitable search target data stored in a database based on input search conditions.

0002

2. Description of the Related Art Generally, expert systems have the ability to search for desired data from a database.

Conventionally, a database is searched using conditions and rules entered by the user, and matching is sequentially attempted with the stored data.
It was determined that the case had not been established.

0004

[Problems to be Solved by the Invention] However, the conventional search function has the following problems.

(1) Since the judgment is based on whether the rule is met or not met, all values that do not meet the standard value but are close to the standard value are determined to be nonconforming and are included in the conditions input by the user (user's intention). Since the ambiguity of the search results is ignored, the search results desired by the user may not always be obtained. For this reason, the user has to repeat the search many times by changing the conditions, which increases the burden on the user.

(2) Since matching is uniformly attempted for the data to be searched, matching is attempted even for items for which other rules clearly do not hold, resulting in a lot of wasteful processing overall. This caused a decrease in processing speed.

SUMMARY OF THE INVENTION In view of these problems, it is an object of the present invention to provide a search device that can reflect ambiguity contained in a user's intention in a search and has excellent processing speed.

[0008]

[Means for Solving the Problems] According to the present invention, in an apparatus that includes a database for storing search target data and searches data matching input search conditions from among the search target data, the following means are provided. It was established.

(1) A first search unit that searches for search target data and selects first search data. This first search unit uses the conditions selected in advance as essential search conditions among the input search conditions, uses fuzzy inference to determine the degree of suitability for the search target data, and searches data for which the degree of suitability is greater than a threshold value. This is selected as the first search data.

(2) A second search unit that performs a predetermined search using fuzzy inference on the first search data using conditions other than the above-mentioned essential search conditions among input search conditions.

[0011]

[Operation] In this device, essential search conditions are set in advance in search condition items according to their importance. When performing a search, the first search section narrows down the search target data based on whether or not the essential search conditions are satisfied at a certain level, and then the second search section searches using other search conditions. This reduces the amount of processing required for clearly non-conforming data. Moreover, fuzzy inference is introduced in the first and second search sections, and suitability is determined based on the degree of suitability, so that ambiguity contained in the user's intention can be appropriately dealt with.

0012

DESCRIPTION OF THE PREFERRED EMBODIMENTS Examples of the present invention will be described below with reference to the drawings. This embodiment is an example of application to a ski tour planning system configured using an expert system.

The expert system is composed of a database, an inference rule base, an inference engine, a man-machine system, etc., and the main functions of the ski tour planning system can be expressed by the functional blocks shown in FIG. In the figure, input functions and output functions are omitted. A database 1 stores predetermined data regarding various ski resorts.

An example of the format of this data is shown in FIG. As shown in Figure 2, this format contains data for each ski resort, such as period (lift operating period), course type (beginner, intermediate, and advanced), travel time and fares by various means of transportation. . As shown in FIGS. 3 to 6, these data are basically expressed by fuzzy numbers. 3 shows the period, FIG. 4 shows the course type, FIG. 5 shows the travel time, and FIG. 6 shows the travel fee. These membership functions are set based on the type of function pattern, median value, distribution value, etc. of the pattern, and membership functions for course types are set using point sequence data.

FIGS. 7 and 8 show the operating procedure of this system. First, the user operates an input unit (not shown) to input essential search items (total budget, lodging costs, scheduled departure date, number of days, skill, transportation means). These conditions are stored in the condition storage section 2. At this time, input of skill is done in the form of inputting the number of beginners, intermediates, and advanced students. If the number of inputs is 0, it will cause a problem in the search, so re-input is requested. In addition, the mode of transportation is entered by selecting the option of airplane, train, bus, or car, and if car is selected, the user is requested to input the number of vehicles.

[0016] Then, the suitability calculation unit 4 calculates the suitability of each ski resort for these items. In order to find the degree of compatibility with respect to the total cost, in this embodiment, the condition calculation unit 3 calculates the total cost from data such as accommodation fees, number of days, means of transportation, etc., and calculates the degree of compatibility between this total cost and the input total budget. The method used is to find the following. Furthermore, when determining the suitability of the period, the tour period is similarly determined from the scheduled departure date and the number of days, and the suitability with the ski resort period is determined. In addition, when calculating the suitability of skill, as shown in Figure 9, the membership function of the course type (indicated by (A) in the figure) and the membership function of the input value (indicated by (B) in the figure) are used. Compare the
For example, a method can be used in which the degree of compatibility is determined from the area of the overlapping part (hatched part) of both graphs. Furthermore, when determining the degree of suitability of transportation means, criteria are used such as, for example, airplanes are suitable for Hokkaido areas, buses are suitable for areas other than Hokkaido areas, etc.

[0017] Then, the evaluation unit 4 evaluates each ski resort based on the degree of suitability obtained in this way, and based on whether or not all the degrees of suitability satisfy the threshold value, the first search result is determined. demand. As a result, ski resorts that do not satisfy important items are excluded from subsequent processing.

[0018] After this, the process moves to the second search. First, the user selects an evaluation method. The evaluation method is the maximum value evaluation method (
It is acceptable if the maximum value of the degree of conformity for each item is within the acceptable range), minimum value evaluation method (it is fine if the minimum value of the degree of conformity for each item is within the acceptable range), average evaluation (the average value of the degree of conformity for each item is acceptable) (It's fine if it's within the range). Based on this input command, the evaluation method selection section 6 selects and sets an evaluation method.

After this, the user inputs search conditions for the second search. The input screen at this time is shown in FIG. Choices 1 to 7 are evaluation items (search condition items), choice 8 is a command to output the inference result (search result), and choice 9 is a command to end the inference operation (search operation). When a search condition item is selected, a request is made to input its importance level or desired value, and based on this input value, the suitability calculation unit 7 and the evaluation unit 8 perform the second
Perform a search. In this second search, the degree of relevance in the first search can also be taken into account for evaluation. The display section displays the search results in an extremely simple format. Furthermore, when the inference result output command is selected, the output method (simple display mode/detailed display mode), output device, and number of outputs of the ski resort are confirmed, and then a predetermined output operation is performed. A simple display screen and a detailed display screen are shown in FIGS. 11 and 12. Thereafter, when a command to end the inference operation is input, a predetermined end operation is performed and the process is ended.

[0020]

[Effects of the Invention] As explained above, the present invention provides the following effects.

(1) Since the first search narrows down the search target based on important search conditions, the second search finally obtains the search results, so clearly unsuitable data, that is, important Processing of data that does not match the search method items is significantly reduced, and processing speed is improved. In particular, when performing a re-search with different search conditions, there is an advantage that the first search data can be used to quickly perform the search, as long as important items among the search conditions remain unchanged.

(2) Since suitability is evaluated based on the degree of suitability using fuzzy inference, ambiguity contained in the user's intention can be reflected in the search results. For this reason,
Desired search results can be obtained without setting extremely strict search conditions, making it easier to set and input search conditions. Moreover, since there is no need to repeat the search many times, the burden on the user is greatly reduced.

[Brief explanation of the drawing]

FIG. 1 is a functional block diagram showing main parts of a ski tour planning system according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of a ski resort data format.

FIG. 3 is a graph showing an example of a membership function expressing a period.

FIG. 4 is a graph showing an example of a membership function expressing course types.

FIG. 5 is a graph showing an example of a membership function expressing travel time.

FIG. 6 is a graph showing an example of a membership function expressing the required travel fee.

FIG. 7 is a flowchart showing a processing procedure in a first search.

FIG. 8 is a flowchart showing a processing procedure in a second search.

FIG. 9 is a graph showing an example of computation of degree of suitability for course types.

FIG. 10 is an explanatory diagram showing an input screen for search conditions, etc. in a second search.

FIG. 11 is an explanatory diagram showing a simple display screen.

FIG. 12 is an explanatory diagram showing a detailed display screen.

[Explanation of symbols]

1...Database 4... Relevance calculation unit in first search 5...Evaluation section in first search 7... Relevance calculation unit in second search 8...Evaluation section in second search

Claims (1)

[Claims] [Claim 1] An apparatus comprising a database for storing search target data, and searching the search target data for data matching input search conditions, wherein the search target data is selected in advance as an essential search condition from among the search conditions. a first search unit that determines the degree of suitability of the search target data by fuzzy inference using conditions, and selects data for which the degree of suitability is equal to or higher than a threshold value as first search data; A search device using fuzzy inference, comprising: a second search unit that performs a predetermined search using fuzzy inference on the first search data using conditions other than the conditions.