JPH10228482A - Taste estimating method, information extracting method and information supplying method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate information which is predicted to be selected by a user in next information retrieval through the use of the change history of the taste of the user. SOLUTION: A means 101 for managing an interesting area being the set of existing access information which the user likes, a means 102 for managing the interesting area being the set of information that are accessed or not accessed yet, in which the user is especially interested at present, a means 103 obtaining and managing change history whether the interested area changes in the interesting area or it changes to outside at the time of an information retrieval trial at every time and a means 104 for selecting accessed information and information which is not accessed yet in accordance with the rate of change history while being coexisted and indicating it to the user as recommendation information are provided.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to human information processing technology, and more particularly, to a system for expressing user preferences in a concrete manner so that a computer system can process the information and presenting information matching the user preferences. The present invention relates to a preference estimation method, an information extraction method, and an information provision method, which are effective when applied to a personal computer.

[0002]

2. Description of the Related Art Conventionally, when information relating to a user's preference is embodied so that it can be processed by a computer, a search history of the user is examined, and the user is determined based on past search results based on information considered to be statistically preferable. Is generally representative of the taste of In this case, the range of the user's preference can be estimated, but how the objects of interest have changed at each point in the past, and as a result, It is impossible to estimate whether or not they have In other words, it can be said that the prior art does not distinguish between “favorite information” and “interesting information”.

[0003]

In the prior art, it is necessary to deal with how the user's preference has changed over time and which information he is currently interested in as a result of the change. Impossible.

[0004] It is an object of the present invention to specify both a favorite area for already-accessed information and information that is particularly interesting at present, and to estimate information expected to be selected by a user in the next information search trial. An object of the present invention is to provide information that matches a user's preference.

[0005]

According to the present invention, there is provided a preference area which is a set of previously accessed information which a user prefers;
Define a region of interest formed by previously accessed or unaccessed information that the user is particularly interested in at that time, and determine whether the region of interest changes within the favorite region in each information search attempt A change history as to whether it changes to the outside is obtained, and the preference of the user is estimated from the change history. Furthermore, according to the ratio of the change history of whether the region of interest changes within the favorite region or changes to the outside, information of previously accessed information and information of non-access are mixed and detected and used as recommendation information. To provide.

[0006]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a system configuration diagram showing one embodiment of the present invention. The system includes a database 10 in which various information is stored,
Information retrieval system 20 for searching for information, a preference estimation system 30 for estimating information that matches the user's preference, and information for presenting information to the user based on the prediction result of the preference estimation system 30 It consists of a presentation system 40.
Here, the preference estimation system 30 is a part according to the present invention, and the preference area management unit 101 that manages a preference area that is a set of previously accessed information that the user prefers, and the user is currently particularly interested. Interest area management means 102 for managing an area of interest, which is a set of information that has been accessed or not accessed, and a preference change history management means 10 for managing a change history of a preference area and an interest area as a change in preference.
3 and the user in the next information search using the preference area information of the preference area management means 101, the preference change history of the preference change history management means 103, and the information other than the preference area obtained through the information search system 20. Is composed of selection information predicting means 104 for predicting information that will be selected.

FIG. 2 shows a flow in the preference estimation system 30 until information to be recommended and presented to a user is determined. Here, it is assumed that the user has performed a k-th search trial in the information search system 20. In addition, it is assumed that a set of search target information existing in the database 10 has been given to the selection information prediction unit 104 via the information search system 20 in advance.

In step 201, it is assumed that the user scores the selected information every time a search is attempted, and inputs his / her satisfaction to the preference estimation system 30. Now, let the ID of the information selected in the k-th search trial be I _k, and let the user's scoring result for this I _{k be} S _k . That is, this scoring result indicates the user's preference. In step 202, the information retrieval system 2
0 indicates that when the k-th search attempt is performed, the selected information ID I _k and the scoring result S _k are assigned to the preference area management unit 10.
1 and sends I _k to the region-of-interest management means 102.

In step 203, the preference area management means 101 determines whether the scoring result S _{k of} the information ID (= I _k ) selected by the user is higher than a score (= S _th ) which is a certain threshold. If the scoring result S _k is higher than S _th , in step 204, the favorite region T _k−1 (= k−
The information ID (= I _k ) selected at the k-th time is added to the information selected at the time of the first search attempt, which is a set of information IDs for which the user's scoring result is higher than S _th. , A new favorite area _Tk . If scoring result is less S _th, in step 205, without the addition of I _k k-
The favorite region T _{k-1 up} to the _{first time} is directly set as a new favorite region T _k . Thereafter, in step 206, the preference area management means 101 sends T _k−1 to the preference change history management means 103 and T _k to the selection information prediction means 104 as the preference area information.

On the other hand, in step 207, the region-of-interest management means 103 determines that the region of interest of the user at the time of the k-th search trial is only the information ID (= I _k ) selected for the k-th search. Form the set F _k of number 1;
This is used as interest area information, and the preference change history management means 103
Send to

In step 208, the preference change history management means 103 sets the interest area F at the time of the k-th search trial.
_It is checked whether or not _k is included in the region of interest T _k-1 up to the (k−1) th search trial. When included (i.e., previously accessed), in step 209, assuming that the user's interest has changed within the preference region, increments the variable P ₁ for counting the number of internal changes. If it is not included (ie, not accessed until the (k-1) th time),
In step 210, assuming that interest has moved to the outside of the preference region, increments the variable P ₀ for counting the number of external changes. After this, step 21
In 1, the preference change history management unit 103 sends the variables P ₀ and P ₁ to the selection information prediction unit 104.

In step 212, the selection information prediction means 104 receives the variables P ₀ and P ₁ as history information from the preference change history management means 103, and determines whether the number ratio of the information included in the favorite area T _k to the information not included is equal to the number. The information (a plurality of pieces) to be presented to the user is determined so that P ₀ : P _1, and the information ID is sent to the information presentation system 40. It should be noted that information not included in the favorite region T _k is _obtained from a set of information to be searched for given through the information search system 20 from T _k.
It is obtained by removing the information contained in.

In step 213, the information presentation system 40 that has received the information ID transmits the corresponding I as recommendation information.
The information of D is presented to the user.

FIG. 3 shows a specific processing example in which a recommended singer to be recommended is determined based on the results of ten singer name searches. Here, the following preconditions are considered.・ The number of singers recommended and presented at a time is five. - the initial value of the preference region T _k is the empty set {}. _-The threshold value S _th for determining whether or not the selected singer likes 7 is
Score 0 (out of 100). When the change of the region of interest is inside, the initial values of the variables P ₁ and P ₀ for counting the number of times of outside are both 0. -It is assumed that a set from singer A to singer Z exists in the singer database.

From the search results shown in FIG. 3, T ₁₀ = ｛A, C,
E｝, P ₁ : P ₀ = 4: 6 = 2: 3. As a prerequisite, since the number of singers presented at one time is five, it is sufficient to select two from singer A, singer C, and singer E and select three other singers from the singer database. Therefore, as examples of presentation each time, [A, C, S, X, Y],
Combinations such as [A, E, O, P, Z] and [C, E, H, S, T] are possible.

[0016]

As described above, according to the present invention, regarding the user's preference, the user's preference area (T _k ), which is a set of previously accessed information, and the user is particularly interested at that time. Region of interest (F _k ), which is a set of previously accessed or unaccessed information, and whether the region of interest changes within the favorite region or changes to the outside at each trial of information retrieval. Find the change history of that,
By using the change history, the user is estimated whether to select information that has not been accessed or information that has been accessed in the next search attempt, thereby estimating the preference.

Therefore, the present invention is a technique capable of estimating information that is currently interesting from the result of a change in the user's preference over time, and is a technique for a type of user who prefers unaccessed new information. In such a case, many types of information outside the favorite area are recommended and presented as new information, and in the case of a type in which the user selects only the already accessed information,
By recommending and presenting a lot of information inside the favorite area,
It is possible to provide information suited to the user's interest at that time.

[Brief description of the drawings]

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

FIG. 2 is a diagram showing a flow of processing until information to be recommended and presented to a user is determined.

FIG. 3 is a diagram showing a specific processing example according to the present invention.

[Explanation of symbols]

 Reference Signs List 10 database 20 information retrieval system 30 preference estimation system 40 information presentation system 101 preference area management means 102 interest area management means 103 preference change history management means 104 selection information prediction means

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Keiichi Ueno 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo Inside Nippon Telegraph and Telephone Corporation

Claims (3)

[Claims] Claims 1. A preference area, which is a set of previously accessed information that a user likes, and an interest area, which is a set of previously accessed information or non-accessed information that the user is particularly interested at that time, are defined. A preference history in which a change history of whether a region of interest changes inside a preference region or changes to the outside in each search trial is obtained, and a preference of a user in a next search trial is estimated based on the change history. Estimation method. 2. An information extracting method to which the preference estimating method according to claim 1 is applied, wherein an interest area has been accessed in accordance with a ratio of a history changing inside a favorite area to a history changing outside. An information extraction method characterized by extracting information and unaccessed information. 3. An information providing method characterized by providing information extracted by the information extracting method according to claim 2 to a user as recommendation information.