JP2012083977A - Retrieval system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a retrieval system that takes into consideration tastes of other users in the same segment or group with a user who targets data of an unspecified field and desires that an element value is within a predetermined range adaptively to the user.SOLUTION: The user who displays data is made to input the element value based upon a determination of the user, and element values of data are prepared based upon element values that many users input so as to hold data having objectivity with respect to various element values. Points are calculated based upon the data so that data matching a taste of the user has high point, and a display based upon the points is made. Further, points are calculated while tastes of other users in the same segment or group with the user are taken into consideration.

Description

  The present invention relates to a retrieval system that retrieves data, ranks the retrieved data, and displays a link to the retrieved data according to the ranking.

  2. Description of the Related Art Internet search systems that search home pages, rank the search results, and display links to the searched home pages according to the ranks are widely used. In addition, it is displayed according to the ranking given to the search result according to the purpose, such as so-called in-site search in which the search target home page is limited to a certain range, or the application software searches for data related to the processing of the software. Things have been done.

In such a search system, the ranking is often based on a uniform standard of the search system. Which data is desired to be displayed at the top differs for each user, but it is not a display that reflects such user preferences.
In order to solve this problem, Patent Document 1 discloses a method in which a user annotates a document and adjusts a search keyword or the like based on the annotation. Further, in Patent Document 2, it is input that whether the web page browsed by the user is compatible / non-compliant, there are many common words with the web page that is adapted, and there are few words that are common with the non-conforming web page. A method of ranking the highest ranking web pages is disclosed.

However, the methods disclosed in Patent Documents 1 and 2 are not sufficient to reflect user preferences in the following points.
The order of data such as a web page to be searched (hereinafter referred to as “search object” is referred to as “data” in this specification) is increased or decreased depending on whether the user has input an annotation or conformity / nonconformity. At this time, the stronger the element, such as “many words in common with the adapted web page”, the higher the ranking.
However, there is a user preference that does not like that a certain element is strong but prefers that a certain element is within a predetermined range without being too strong or too weak. For example, red wine prefers light body or medium body rather than full body, and manga prefers moderate romance expression.
As described above, it has not been possible to cope with the user's preference that the element value prefers to be within a predetermined range.

In order to cope with the user's preference that the element value prefers to be within a predetermined range, the following problem must be solved.
Various element values can be considered for the data. For example, element values such as the content level of romance expression, the content level of violence expression, and the age of the main character can be considered for comics. It is not easy to prepare all of these element values for all data to be searched. If the element value has an objective definition that can be measured, such as the degree of sake for sake, the measured value can be used, but the content of love expressions for comics can depend on the subjectivity of the user It is difficult to prepare objective element values.
For this reason, a search system corresponding to the user's preference that prefers that the element value is within a predetermined range for data in an unspecified field has not been provided.

JP 2010-146595 A JP 2007-148720 A

  The problem to be solved is that a search system corresponding to the user's preference that prefers that the element value is within a predetermined range for data in an unspecified field has not been provided.

  By letting the user who displayed the data input the element value based on the user's judgment and preparing the element value of the data based on the element value input by many users, the element for the data in unspecified fields Provided is a search system corresponding to a user's preference that prefers a value to be within a predetermined range.

といった関数の値とすればよい。ここで、ｆはユーザ嗜好合致度、ｕはユーザ要素値、ｄはデータ要素値を示し、ｃは定数である。 The search system of the present invention
In a search system for searching for data, ranking the searched data, and displaying a link to the searched data according to the rank,
The ranking is higher as the score calculated for each data is higher,
Data element value processing means for holding one or more element values (data element values) corresponding to each of the data;
User element value processing means for holding the element value (user element value) corresponding to a user (user) of the search system;
User preference match degree calculating means for reading the data element value and the user element value and calculating a user preference match degree based on a calculation formula such that the smaller the difference is, the larger the value is.
The higher the user preference match value, the higher the score.
The user element value represents the user's preference. The data element value represents an element of each data. Here, the “calculation formula that increases as the difference is smaller” between the user element value and the data element value may be a monotonously decreasing function of the difference between the user element value and the data element value. For example,

The value of a function such as Here, f indicates a user preference match degree, u indicates a user element value, d indicates a data element value, and c is a constant.

The search system of the present invention
User evaluation input means for allowing the user to input the element value based on the user's judgment,
The data element value processing means calculates and holds the data element value based on the element value input by the user evaluation input means.
The data element value depends on the subjective judgment of the user, and it is often difficult to obtain an objective value. Therefore, the user is allowed to input an element value based on the user's judgment, and the data element value is calculated based on many user inputs, thereby obtaining an objective data element value. For example, an average value of inputs by many users can be used as the data element value. Of course, instead of a simple average of the user's input, a weighted average that causes the weight of recent input to be large is used. In addition to the average value, the mode (media) or median (median) is used instead of the average value. It is also possible to use the input of the search system operator.
The user evaluation input means is not necessarily used at the time when the user displays the data, and the user may start the user evaluation input means after displaying the data (for example, several days later). For example, with regard to a user evaluation for a restaurant, it is conceivable that after displaying data, the user eats at the restaurant and then inputs the user evaluation.

ここで、ｄiは、ユーザが表示したｉ番目のデータの表示時点におけるデータ要素値である。 The search system of the present invention
The user element value processing means calculates and holds the user element value based on the data element value of the data displayed by the user via the link to the data.
The user element value represents the preference of the user, and in many cases, the user can set the value. For example, a means for a user to input a user element value may be provided, and the user may input using the means.
However, it is considered that there are users who prefer to use values estimated by the system instead of inputting them themselves, for example, because there are many types of element values and there are elements that the user does not have a strong preference.
For such a user, the user element value can be estimated using the data element value of the data displayed by the user via a link to the data. This is because it is estimated that the user is interested in the displayed data. For example, as described below, the user element value can be obtained by averaging the data element values of the displayed data.

Here, di is a data element value at the time of display of the i-th data displayed by the user.

ここで、ｗ_iは、ユーザの入力に基づいて定まるウエイトであり、例えば、「完全に嗜好と合っている」ものについては１、「おおむね嗜好と合っている」ものについては０．５、「嗜好と合っていない」ものについては０というように定めることができる。または、満足度を例えば１００点満点の数値で入力させ、その数値に基づいて定めることができる。 The search system of the present invention
A satisfaction level input means for allowing the user to input a value (satisfaction level) as to whether or not the data meets the user's preference;
The user element value processing means calculates and holds the user element value based on a value obtained by weighting the data element value based on the satisfaction degree.
When the user displays the data, it cannot be said that the data matches the user's preference. Data that does not match preferences may be displayed, and even data that matches preferences may be completely matched data or partially matched data.
Therefore, the user is made to input whether or not the displayed data matches the user's preference, and to what extent, the user element value is calculated based on the weighted value based on the input. It is also possible. Specifically, the following formula is used instead of the above formula 3.

Here, w _i is a weight determined based on the user's input, for example, 1 for “completely fits the preference”, 0.5 for “mostly matches the preference”, “ For items that do not match preferences, it can be set to 0. Alternatively, the degree of satisfaction can be input based on a numerical value of 100 points, for example, and determined based on the numerical value.

ここで、ｗ_jは同一セグメントに属する他のユーザの満足度に基づくウエイトを表す。関数ｇの値が、他のユーザの満足度に基づくウエイトの平均値であり、セグメント得点となる。ｎは、同一セグメントに属する他ユーザの人数である。 The search system of the present invention
Users are categorized by their attributes,
A segment score calculating means for calculating a segment score based on a calculation formula such that the larger the satisfaction value of other users belonging to the same segment as the user currently used is, the larger the value is;
The higher the segment score value, the higher the score.
For a single user, particularly at the beginning of using a search system, the accuracy of user element values may not be sufficient. Therefore, this lack of accuracy can be compensated by using the satisfaction of other users belonging to the same segment as the user. Here, the “segment” is a classification of users determined by individual user attributes such as age group, gender, and residence. Users who belong to the same segment are considered to have similar preferences. For example, the following formula is calculated.

Here, w _j represents a weight based on the satisfaction of other users belonging to the same segment. The value of the function g is an average value of weights based on satisfaction of other users, and becomes a segment score. n is the number of other users belonging to the same segment.

ここで、ｒ（ｉ，ｊ）はユーザｉの要素値とユーザｊの要素値との類似度、ｆ_ｋは、ｋ番目の要素値についてのユーザ嗜好合致度を表す関数である。
このｒ（ｉ，ｊ）を用いて、例えば以下の式により同族得点を計算することができる。類似度が所定の値以上であるような他のユーザに係る満足度を用いる場合には、ユーザ０に係る同族得点は、

となる。ここで、Ｒは「所定の値」、ｎはｒ（０，ｊ）≧Ｒを満たすようなユーザｊの数である。
類似度に基づくウエイトをつけたユーザ嗜好合致度を用いる場合には、ユーザ０に係る同族得点は、

となる。
むろん、ユーザ嗜好合致度、セグメント得点及び同族得点にウエイトをつけ、

に基づいて得点を計算することができる。ここで、ｓはセグメント得点の、ｔは同族得点のウエイトを表す。 The search system of the present invention
A degree-of-similarity calculating means for calculating the degree of similarity of the user element value relating to the user and other users;
A family score calculation that calculates a family score based on a formula that increases as the degree of satisfaction of other users whose degree of similarity with the user currently in use is greater than or equal to a predetermined value With means,
The higher the score of the family, the higher the score.
Here, the “similarity” increases as the preferences of two users are similar. For example, the similarity r can be calculated as follows using the function f representing the user preference matching degree.

Here, r (i, j) is a similarity between the element value of the user i and the element value of the user j, and f _k is a function representing a user preference matching degree for the k-th element value.
Using this r (i, j), for example, a family score can be calculated by the following equation. When using satisfaction related to other users whose similarity is equal to or greater than a predetermined value, the relative score for user 0 is

It becomes. Here, R is a “predetermined value”, and n is the number of users j that satisfy r (0, j) ≧ R.
When using a user preference match with a weight based on similarity, the family score for user 0 is

It becomes.
Of course, weighting user preference match, segment score and family score,

Scores can be calculated based on Here, s represents the weight of the segment score, and t represents the weight of the family score.

In the above, the means by which the present invention solves the problem for one element value has been described. Of course, a score can be determined by summing up a plurality of element values. Further, not only the element values that are preferred to be within a predetermined range but also the element values that are preferred as the value is larger can be used to determine the score.
Note that the mathematical formulas used to describe the means for solving the problem are merely examples, and as long as the user preference match is a monotonically decreasing function of the difference between the user element value and the data element value, various formulas can be used as shown in the following examples. Deformation is possible.

  There is provided a search system corresponding to a user's preference that prefers that an element value is within a predetermined range for data in an unspecified field. Moreover, the preference of the other user of the same segment or the same family can be considered.

FIG. 1 is a diagram illustrating an example of a configuration of a search system. FIG. 2 is a diagram illustrating an example of data element values. FIG. 3 is a diagram illustrating an example of user element values. FIG. 4 is a diagram illustrating an example of the satisfaction value. FIG. 5 is a diagram illustrating an example of processing at the time of search. FIG. 6 is a diagram illustrating an example of processing by the user preference matching degree calculation unit. FIG. 7 is a diagram illustrating an example of a screen displaying a link list and data. FIG. 8 is a diagram illustrating an example of a screen of the satisfaction degree input unit. FIG. 9 is a diagram showing an example of a screen of user evaluation input means. FIG. 10 is a diagram showing an example of processing by the segment score calculation means. FIG. 11 is a diagram showing an example of processing by the family score calculation means.

  Examples of the present invention will be described below. The configuration of the search system, the data format, and the basic flow at the time of search are common to all the embodiments.

(Configuration of search system)
FIG. 1 is a diagram illustrating an example of a configuration of a search system. The search means 1 searches the database 6 and extracts data that matches the search conditions. The link list display means 2 calculates a score for the retrieved data and displays a link to the data in descending order of the score. When the user selects one of the displayed links (for example, clicks with the mouse), the data display means 3 displays the data.
For the calculation of the score by the link list display means 2, the user preference matching degree calculation means 21, the segment score calculation means 22 and the family score calculation means 23 can be used. The family score calculation means 23 uses the similarity calculation means 24 to calculate the degree of similarity between the preferences of two users.
The user preference match degree calculation means calculates a user preference match degree by comparing a data element value 51 that is an element value of each data and a user element value 52 that represents the user's preference by an element value. The data element value 51 is updated by the data element value processing means 41. The user element value 52 is updated by the user element value processing means 42.
The segment score calculating means 22 and the affiliated score calculating means 23 calculate scores based on satisfaction values of other users who are in a predetermined relationship (same segment or similar user preference) with the user currently in use.
The data display unit 3 passes information indicating which data the user has displayed to the user element value processing unit 42. The data display means 3 is a satisfaction input means 31 for the user to input whether or not the displayed data matches the user's preference, and the user inputs an element value for the user to evaluate the displayed data. The user evaluation input means 32 can be called by the user. When the satisfaction is input by the user, the satisfaction input unit 31 and the user evaluation input unit 32 store the value as the satisfaction value 53.
The data element value processing means 42 updates the data element value 51 using the information passed from the user evaluation input means 32. The user element value processing means 42 updates the user element value 52 using information passed from the data display means 3, the satisfaction degree input means 31 and the user evaluation input means 32.

(data form)
FIG. 2 is a diagram illustrating an example of data element values. Each element value is held for each data (D1, D2, and D3). Also, the evaluation number for the data (the number of inputs by the user evaluation input means 32) is held. The element value is an average value of values input by the user evaluation input unit 32.
FIG. 3 is a diagram illustrating an example of user element values. For each user (U0, U1,... U5), segment information, weights, and element values are held. The segment information is used by the segment score calculation means 22 to determine a segment, and is age and sex in this embodiment. In addition to this, other information such as a residence and a hobby may be included. The element value is the same as the data element value but retains not only the mean but also the standard deviation. Here, the standard deviation is calculated using the information passed from the data display means 3, satisfaction input means 31 and user evaluation input means 32 for the user as a population. The weight is the total sum of the weights of the information passed from the data display means 3, the satisfaction degree input means 31, and the user evaluation input means 32. One user may search for data having completely different elements (for example, restaurants and comics). In that case, the weight and element value for the different data of each element for each user should be held, or the data of the user element value should be held separately for the different data of each element. . In the following, a search for comics on the subject of sports, which is data processed by one set of element values, will be described.
FIG. 4 is a diagram illustrating an example of the satisfaction value. It is stored which user has entered what satisfaction for which data.

(Basic flow when searching)
FIG. 5 is a diagram illustrating an example of processing at the time of search. In a search for comics related to sports, the user inputs “search subject is baseball” as a search condition and executes the search.
The search means 1 extracts data from the database 6 such that the subject is baseball.
The link list display means 2 ranks the extracted data and displays a link list according to the rank. The link list display means 2 calls the user preference match degree calculation means 21, the segment score calculation means 22 and the family score calculation means 23 for ranking, obtains the user preference match degree, the segment score, and the family score, and weights them. Calculate the score with the, and determine the ranking. The segment score calculation means 22 and the family score calculation means 23 do not necessarily need to be called, and the rank may be determined based only on the user preference match degree.

  In this embodiment, the order is determined based only on the user preference match degree.

ここでσは標準偏差であり、標準偏差で規格化することによってユーザが各要素値を重視する度合いを反映する。ただし、標準偏差が０．１未満の場合は０．１とする。例えばユーザが表示したデータが１つだけの場合に標準偏差が０となり、式の計算が不可能になることを防止するためである。定数の値を４０としているのは、要素値が１〜５の５段階で評価されユーザ要素値ｕとデータ要素値ｄとは最大で４の差があり、規格化に用いる標準偏差が０．１の場合にも嗜好合致度の値を正または０とするためである。ただし、定数の値は順位を変動させないので、任意の値でよい。
図２に示すデータについて、ユーザ嗜好合致度を計算すると、データＤ１のユーザ嗜好合致度はおよそ１１９．１、データＤ２のユーザ嗜好合致度はおよそ１１８．１、データＤ３のユーザ嗜好合致度はおよそ１１３．２となる。
リンク一覧表示手段２は、ユーザ嗜好合致度を得点とし、その順、すなわち、データＤ１、Ｄ２、Ｄ３の順にリンク一覧を表示する。
なお、ユーザ要素値５２の平均値及び標準偏差に替えて、図示しない手段によってユーザに検索したいデータ要素値の範囲を指定させることもできる。このようにすると、例えば、ユーザが自身の嗜好とは異なるデータを意図的に検索することができる。 The link list display means 2 calls the user preference match degree calculation means 21 for ranking. FIG. 6 is a diagram illustrating an example of processing by the user preference matching degree calculation unit. The user preference matching degree calculation means 21 calculates and outputs the preference matching degree of each data. The data preference coincidence is the sum of the element value preference coincidence.
In this embodiment, the search is for a comic about sports. Element value 1 is a sense of urgency about the outcome of the game, element value 2 is the degree of expression of the player's guts, and element value 3 is the degree of love expression. is there.

The user “U0” in FIG. 3 performs a search. In FIG. 3, the value of each element value related to the user U0 is determined based on the data element value of the data displayed by the user U0. The user U0 has an element value 1 (a sense of urgency about winning or losing the game) of data around 3.3, an element value 2 (degree of expression of the player's guts) around 3.2, an element value 3 (romance expression) Prefers data of around 2.8.
Here, the standard deviation value of the element value 3 is larger than the element value 1 and the element value 2. This means that the variation of the element value 3 in the data displayed by the user U0 is large, and it is suggested that the user U0 does not attach importance to the element value 3 (degree of love expression).
The preference matching degree in step S02 is calculated according to the following equation.

Here, σ is a standard deviation, and reflects the degree to which the user attaches importance to each element value by normalizing with the standard deviation. However, when the standard deviation is less than 0.1, it is set to 0.1. For example, this is to prevent the standard deviation from becoming zero when the user displays only one piece of data, making it impossible to calculate the formula. The constant value is set to 40 because the user element value u and the data element value d have a maximum difference of 4 and the standard deviation used for normalization is 0. This is because the preference match value is set to positive or 0 even in the case of 1. However, the value of the constant does not change the order, and may be an arbitrary value.
When the user preference match degree is calculated for the data shown in FIG. 2, the user preference match degree of the data D1 is approximately 119.1, the user preference match degree of the data D2 is approximately 118.1, and the user preference match degree of the data D3 is approximately 113.2.
The link list display means 2 uses the user preference match score as a score, and displays the link list in that order, that is, in the order of data D1, D2, and D3.
Note that, instead of the average value and the standard deviation of the user element value 52, the range of the data element value that the user wants to search can be specified by means not shown. In this way, for example, the user can intentionally search for data different from his / her preference.

ここで、ｗ，ｕ，σはそれぞれウエイト、平均、標準偏差であり、更新後の値は「’」を付したものである。また、ｗ_０は渡されたデータに係るウエイトであり、ここでは１．０とする。ｄは表示されたデータの要素値であり、データ要素値４１に保持されている。 FIG. 7 is a diagram illustrating an example of a screen displaying a link list and data. The screen shown in FIG. 7A is displayed by the link list display means 2.
When the user clicks the link to the data D1, the data display unit 3 displays the data D1. This display is the screen of FIG. A simple scoring button 71 and a detailed evaluation button 72 are provided on the screen.
The data display means 3 passes to the user element value processing means 42 that the data D1 is displayed by the user U0. The user element value processing means 42 updates the weight and element value relating to the user U0 of the user element value 52 based on the passed data.

Here, w, u, and σ are weight, average, and standard deviation, respectively, and the updated values are given with “′”. Further, w ₀ is a weight related to the passed data, and is assumed here to be 1.0. d is an element value of the displayed data, and is held in the data element value 41.

FIG. 8 is a diagram illustrating an example of a screen of the satisfaction degree input unit. When the user clicks the simple scoring button 71, the screen in FIG. 8A is displayed. The user inputs whether or not the displayed data matches his / her preference and transmits it.
The satisfaction input means 31 sets the satisfaction value as 100 for the “very good” data, 75 for the “good” data, and 25 for the “not good” data. Save to 53. Note that the screen shown in FIG. 8B may be displayed instead of the screen shown in FIG. 8A, and the user may input a numerical value of satisfaction.
The satisfaction degree input unit 31 passes the suitability input by the user to the user element value processing unit 42. The user element value processing means 42 updates the weight and element value relating to the user U0 of the user element value 52 based on the passed data. The weight is set to 2.0 for “very good” data, the weight is set to 1.5 for “first time fit” data, and the weight is set to 0 for data “does not match”. Since the weight is already processed as 1 when the data is displayed, the user element value processing means 42 sets w0 to “1.0” for the “very good” data, For the “data”, w0 is set to “0.5”, and for the “not matched” data, w0 is set to “−1.0” and the calculation of Expression 11 is performed. Thereby, the weight of the data processed with the weight of 1.0 is corrected to 2.0, 1.5, or 0.
Here, the reason why the weight of the data “matched well” is set to 2.0 is to increase the weight of the data for which the user has input the satisfaction level as compared with the data simply displaying the search result.

ここで、ｎは評価数、ｄは要素値であり更新後の値は「’」を付したものである。また、ｄ_０はユーザが入力した値である。
ユーザ要素値処理手段４２は、ウエイトを３．０とし、ｄの値としてユーザが入力した値を用いて、数１１の計算を行う。ウエイトを３．０とするのは、ユーザが評価を入力したデータについては、単に検索結果を表示したデータ及び満足度のみを入力したデータよりもウエイトを大きくするためである。 FIG. 9 is a diagram showing an example of a screen of user evaluation input means. When the user clicks the detailed evaluation button 72, this screen is displayed. The user inputs a self-evaluation by dragging a black circle for each element value, inputs a numerical value of satisfaction, and transmits it.
The user evaluation input means 32 stores the satisfaction value input by the user in the satisfaction value 53.
The user evaluation input means 32 passes the element value input by the user to the data element value processing means 41 and the user element value processing means 42.
The data element value processing means 41 increases the evaluation number of the data element value 41 by 1, and updates the element value as shown in the following equation.

Here, n is the number of evaluations, d is an element value, and the updated value is given with “′”. D ₀ is a value input by the user.
The user element value processing means 42 calculates Equation 11 using a weight of 3.0 and a value input by the user as the value of d. The reason why the weight is set to 3.0 is to increase the weight of the data for which the user inputs the evaluation as compared with the data for which the search result is simply displayed and the data for which only the satisfaction level is input.

The above processing provides a search system that estimates user preferences based on data displayed by the user, satisfaction input by the user, and evaluation input by the user, and displays data that matches the user preference at the top. The
The satisfaction degree input means 31 and the user evaluation input means 32 are not necessarily called by the data display means. The satisfaction degree input means 31 or the user evaluation input means 32 may be called by another method not shown in FIG. For example, the user can call and input the satisfaction degree input means 31 or the user evaluation input means 32 after reading the comic of the search result.
For each data, the user may designate “read” or “unread”. The already read comics can be prevented from being displayed as search results. Further, the data for which the satisfaction level is input by the satisfaction level input unit 31 and the user evaluation input unit 32 can be processed as “read”. If the search target is not a comic, it is possible that there is no distinction between “read” and “unread”, but in any case, a method is provided in which the user designates “not included in search results”.

  In this embodiment, the ranking is determined in consideration of the segment score and the family score in addition to the user preference matching degree.

  The link list display means 2 calls the user preference matching degree calculation means 21, the segment score calculation means 22, and the family score calculation means 23 for ranking. The processing of the user preference match degree calculation means is the same as in the first embodiment.

FIG. 10 is a diagram showing an example of processing by the segment score calculation means. The segment score calculation means 22 lists other users of the same segment as the user U0. Since U0 is a 35-year-old woman, a woman in her 30s is assumed to be another user in the same segment. U3 and U5 are listed.
The segment score calculation means 22 calculates a weight based on the satisfaction levels of U3 and U5, and outputs the average value as a segment score. Here, in calculating the weight, 50 is subtracted from the satisfaction value. It is based on the idea that it is a negative evaluation that the satisfaction level (0 to 100 points) scored with a maximum score of 100 is less than 50 points.

ここで、Σは全ての要素値についての和であり、ｕは各ユーザのユーザ要素値、ｕ_０はユーザＵ０の要素値である。
同族得点計算手段２３は、Ｕ１及びＵ３の満足度に基づくウエイトを計算し、その平均値を同族得点として出力する。ここで、ウエイトの計算にあたり、満足度値から５０を減算している。１００点満点で採点される（０〜１００点）の満足度が５０点未満であることは否定的な評価であるとの考えによるものである。 FIG. 11 is a diagram showing an example of processing by the family score calculation means. The family score calculation means 23 lists the user U0 and other users in the family. The similarity calculation means 24 is used for this listing. U1 and U3 are listed as users satisfying the following expression.

Here, Σ is the sum for all element values, u is the user element value for each user, and u ₀ is the element value for user U ₀ .
The family score calculation means 23 calculates a weight based on the satisfaction levels of U1 and U3, and outputs the average value as a family score. Here, in calculating the weight, 50 is subtracted from the satisfaction value. It is based on the idea that it is a negative evaluation that the satisfaction level (0 to 100 points) scored with a maximum score of 100 is less than 50 points.

  The link list display means 2 uses a score obtained by adding a weight of 1.0 for the user preference match, 0.004 for the segment score, and 0.003 for the family score, and links to each data in descending order of score. Is displayed. In addition, since the segment score and the family score have a value of 0 to 100 and are larger than the user preference match degree, the segment score and the family score are adjusted.

  The designation of “read” or “unread” by the user has a significant effect in this embodiment. This is because other users are highly satisfied and data related to comics that are not read by the user are displayed at the top.

  This is a search system corresponding to user's preference that prefers that the element value is within a predetermined range for unspecified field data, and can be expected to be used by many users.

DESCRIPTION OF SYMBOLS 1 Search means 2 Link list display means 21 User preference coincidence calculation means 22 Segment score calculation means 23 Family score calculation means 24 Similarity calculation means 3 Data display means 31 Satisfaction input means 32 User evaluation input means 41 Data element value processing means 41 42 User element value processing means 51 Data element value 52 User element value 53 Satisfaction value 6 Database 71 Simple scoring button 72 Detailed evaluation button

Claims (6)

In a search system for searching for data, ranking the searched data, and displaying a link to the searched data according to the rank,
The ranking is higher as the score calculated for each data is higher,
Data element value processing means for holding one or more element values (data element values) corresponding to each of the data;
User element value processing means for holding the element value (user element value) corresponding to a user (user) of the search system;
User preference match degree calculating means for reading the data element value and the user element value and calculating a user preference match degree based on a calculation formula such that the smaller the difference is, the larger the value is.
The search system, wherein the score is higher as the user preference match value is larger. User evaluation input means for allowing the user to input the element value based on the user's judgment,
The search system according to claim 1, wherein the data element value processing means calculates and holds the data element value based on the element value input by the user evaluation input means.   3. The user element value processing means calculates and holds the user element value based on the data element value of data displayed by a user via a link to the data. The search system described in. A satisfaction level input means for allowing the user to input a value (satisfaction level) as to whether or not the data meets the user's preference;
The search according to claim 3, wherein the user element value processing means calculates and holds the user element value based on a value obtained by weighting the data element value based on the satisfaction degree. system. Users are categorized by their attributes,
A segment score calculating means for calculating a segment score based on a calculation formula such that the larger the satisfaction value of other users belonging to the same segment as the user currently used is, the larger the value is;
The search system according to claim 4, wherein the higher the segment score, the higher the score. A degree-of-similarity calculating means for calculating the degree of similarity of the user element value relating to the user and other users;
A family score calculation that calculates a family score based on a formula that increases as the degree of satisfaction of other users whose degree of similarity with the user currently in use is greater than or equal to a predetermined value With means,
The search system according to claim 4, wherein the score is higher as the value of the family score is higher.

Images