JP6178366B2 - Information processing system, information processing method, and program - Google Patents

Description

  The present invention relates to an information processing system, an information processing method, and a program.

  2. Description of the Related Art Conventionally, a technique for learning a user's interest in a content on the client terminal side based on content acquired from a predetermined web server via the Internet is already known. Interest learning can also be performed by acquiring an RSS (Real Simple Syndication) feed as a data format used for providing users with content such as news, images, moving images, audio, and posting to SNS. The RSS feed is read without accessing the original content by an RSS reader that provides information about the content such as the content name and summary.

  For example, Patent Document 1 discloses a technique for providing a content from a server to a client terminal by extracting a general highly topical article from news information distributed within a certain period designated by a user. A news information presentation device to be presented to a person is disclosed. The news information presentation apparatus according to Patent Document 1 includes means 2 for dividing news information distributed in the form of e-mail, magazines, etc. into articles, and means 3 for selecting useful information from the divided articles. And means 4 for collecting the selected articles and reconstructing them into news information in a format to be presented to the user.

JP 2012-203683 A

  By the way, in the technology for providing content such as news information disclosed in Patent Document 1, for example, all content is learned regardless of whether the user reads the provided content or not. However, the user only reads some contents, and most contents are learned although the contents are not read by the user. Since a large amount of unread content that cannot be read by the user is learned, the amount of calculation and the amount of information to be learned become enormous, and the load on the apparatus becomes excessive.

  The present invention has been made in view of such a situation, and an object thereof is to reduce an apparatus load in content learning.

  In order to solve the above-described problem, an information processing system according to the present invention includes a management device that manages the appearance frequency of a feature amount included in acquired content and an information processing device that learns a user's interest in the content. The information processing system is connected to the information processing system via the appearance frequency calculating means for calculating the appearance frequency of the feature amount included in the acquired content, and the feature amount and appearance frequency calculating means for calculating the feature amount included in the acquired content. An appearance frequency transmitting means for transmitting the appearance frequency to the information processing apparatus, the information processing apparatus receiving a feature amount included in the content and the appearance frequency calculated by the appearance frequency calculating means, and a receiving means A first storage means for storing the appearance frequency received in accordance with the feature amount included in the content, and the code received by the receiving means. Based on the second storage means for storing, as learning information, the probability that the feature included in the content is included in the content for which the predetermined processing has been executed, and the feature amount included in the content for which the predetermined processing has been executed. Interest for evaluating the degree of interest in the feature quantity included in the acquired content based on the learning means for updating the learning information stored in the second storage means, and the appearance frequency and the learning information stored in the second storage means And a degree evaluation means.

  According to the present invention, it is possible to reduce an apparatus load in content learning.

1 is an overall configuration diagram of an information processing system in an embodiment of the present invention. It is a schematic diagram which shows the outline of the appearance frequency table in embodiment of this invention. It is a flowchart which shows the information processing procedure in embodiment of this invention. It is a whole block diagram of the information processing system which concerns on the modification of this invention. It is a schematic diagram explaining the example of word-based naive Bayes.

  An information processing system according to an embodiment of the present invention will be described below with reference to the drawings. In addition, in each figure, the same code | symbol is attached | subjected to the part which is the same or it corresponds, The duplication description is simplified thru | or abbreviate | omitted suitably. The embodiment described below is the best mode of the present invention, and does not limit the scope of the claims according to the present invention.

  The configuration of the information processing system in the present embodiment will be described with reference to FIG. The information processing system in the present embodiment is configured by connecting a management device 10 and an information processing device 20 via a network. The network is composed of, for example, a WAN or a LAN, and may be wireless or wired.

  The management device 10 manages the appearance frequency of the feature amount included in the acquired content. The management device 10 acquires content from the content server 30. The content includes, for example, HTML content itself, RSS, which is an XML-based format that describes metadata such as website headings and summaries, program content provided by TV broadcasting, and video on demand. Includes provided video content and the like. This embodiment will be described using RSS as content.

  The feature amount included in the content may include other attributes that can be classified, such as color and loudness, in addition to the word information constituting the summary included in the RSS and the genre information included in the program content. Good. The appearance frequency of the feature amount refers to the degree to which each word is included in a summary or the like in terms of word information included in RSS, for example.

  The information processing device 20 refers to a processing device for learning the user's interest in the content, for example, a personal computer, a smartphone, and a television for processing the content. In the present embodiment, the “process” includes all processes in which other contents such as browsing, confirmation, viewing, recording, copying, recording, saving, deleting, transmission / reception, and transfer of contents are executed in a form desired by the user. Is included. By combining these processes, the user's interest in the content is estimated.

  The management apparatus 10 includes a content acquisition unit 11, a content management unit 12, a feature amount totaling unit 13, an appearance frequency table 14, an appearance frequency transmission unit 15, and a communication unit 16.

  The content acquisition unit 11 acquires content from the content server 30, and the function is realized by, for example, a network interface. In the present embodiment, the content acquisition unit 11 functions as an RSS feed reception unit that receives an RSS feed.

  The content management unit 12 functions as a content management unit such as storing, deleting, organizing, and transferring the acquired content to another functional unit. In the present embodiment, the content management unit 12 functions as an RSS feed cache that stores an RSS feed. The function of the content management unit 12 is realized by a storage device such as an HDD.

  The feature amount totaling unit 13 functions as an appearance frequency calculating unit that divides an article such as a summary included in the RSS feed into word information and calculates appearance frequency information of the word information included in the summary. For example, morphological analysis may be used as a method for specifying word information from the RSS feed. The function of the feature amount totaling unit 13 is realized by, for example, a CPU or the like reading a predetermined program for realizing the function from the ROM or the like to the RAM or the like and executing it.

  The appearance frequency table 14 is a storage device such as an HDD for storing the appearance frequency information of word information.

  The appearance frequency transmitting unit 15 functions as an appearance frequency transmitting unit that transmits the feature amount included in the content and the appearance frequency calculated by the feature amount totaling unit 13 to the information processing apparatus 20. For example, the function is realized by a network interface or the like. Is done. In the present embodiment, the appearance frequency transmission unit 15 transmits the word information and the appearance frequency information to the information processing apparatus 20.

  The communication unit 16 is a communication unit that transmits various contents to the information processing apparatus 20, and the function is realized by, for example, a network interface.

  The information processing apparatus 20 includes a content receiving unit 12, an appearance frequency receiving unit 22, a content storage unit 23, an appearance frequency storage unit 24, a learning unit 25, a learning information storage unit 26, and an interest degree evaluation unit 27. The input unit 28 and the output unit 29 are included.

  The content receiving unit 12 receives the content transmitted from the management apparatus 10, and its function is realized by, for example, a network interface.

  The appearance frequency receiving unit 22 is a receiving unit that receives the feature amount included in the content and the appearance frequency calculated by the feature amount totaling unit 13, and its function is realized by, for example, a network interface.

  The content storage unit 23 is a storage device such as an HDD that stores the content received from the management device 10 by the content reception unit 21. In the present embodiment, the content storage unit 23 stores the received RSS feed.

  The appearance frequency storage unit 24 is a storage device such as an HDD that functions as a first storage unit that stores the appearance frequency received by the appearance frequency reception unit 22 in association with the feature amount included in the content. In the present embodiment, the appearance frequency storage unit 24 stores the appearance frequency of word information.

  The learning unit 25 functions as a learning unit that updates the learning information stored in the learning information storage unit 26 based on the feature amount included in the content that has been subjected to the predetermined processing. The details of the learning unit 25 will be described later. For example, the word information included in the content on which the predetermined process has been executed may be a process such as weighting the word information, for example, giving 1 count. The function of the learning unit 25 is executed by a CPU or the like included in the information processing apparatus 20 by reading a predetermined program that realizes the function from a storage unit such as a ROM to a temporary storage unit such as a RAM and executing it. Realized.

  The learning information storage unit 26 is a storage device such as an HDD that functions as a second storage unit that stores, as learning information, the probability that each feature amount is included in the content that has been subjected to predetermined processing. In the present embodiment, the learning information storage unit 26 functions as a database that stores the probability included in the RSS feed that has been browsed, for example, for each word information. In the present embodiment, the appearance frequency storage unit 24 is defined as the first storage unit, and the learning information storage unit 26 is defined as the second storage unit, which is formal. It is not a definition of superiority.

  The degree-of-interest evaluation unit 27 functions as an degree-of-interest evaluation unit that evaluates the degree of interest with respect to the feature amount included in the acquired content based on the appearance frequency and the learning information stored in the learning information storage unit 26. Then, the degree-of-interest evaluation unit 27 determines the probability that the word information is included in the content that has been subjected to the predetermined process as the learning information stored in the learning information storage unit 26, the content that has been subjected to the predetermined process, and the predetermined content. The degree of interest in the feature amount included in the acquired content is evaluated based on the difference from the probability that the word information is included in the content that has not been processed. Although details of the interest level evaluation unit 27 will be described later, in this embodiment, naive Bayes is used as one of methods for realizing a function for evaluating the interest level.

  The input unit 28 is an input device such as a keyboard or a mouse that functions as an input receiving unit that receives an input / operation on the information processing apparatus 20 by the user. Upon receiving a user input using the input unit 28 for processing a certain content, the learning unit 25 determines whether or not a predetermined process has been executed for the content.

  The output unit 29 functions as, for example, a display unit such as a display that displays content, an RSS feed, or the like, and a voice output unit such as a speaker that outputs audio data included in the content.

  Note that the management apparatus 10 may include a conversion unit that converts the feature amount included in the acquired content into predetermined identification information. Specifically, the appearance frequency table may have a function as conversion means. Then, the appearance frequency transmission unit 15 receives the identification information converted from the feature amount included in the content in the appearance frequency table and the appearance frequency of the feature amount included in the content calculated by the feature amount totaling unit 13. When the appearance frequency receiving unit 22 receives the identification information converted from the feature amount included in the content and the appearance frequency calculated by the feature amount totaling unit 13, the appearance frequency storage unit 24 displays the appearance frequency. The appearance frequency received by the receiving unit 22 is stored in association with the identification information converted from the feature amount included in the content.

  In addition, the appearance frequency transmission unit 15 in the management device 10 may transmit, to the information processing device 20, a feature amount that is predetermined as a feature amount included in the content. The predetermined feature amount means, for example, an important word or the like obtained by narrowing word information included in each content by using, for example, Team Frequency (TF) / Inverse Document Frequency (IDF).

[Conventional problems of interest evaluation using naive Bayes]
As an example in which the present embodiment operates particularly effectively, there is an interest degree evaluation using naive Bayes. This problem will be described with reference to FIG. Here, first, as shown in [1], a summary in the RSS feed that “XX soccer representatives will challenge the World Cup △△ tournament XX competition with □□□ representatives at the stadium” (text) ) Is a collection of words. In other words, as shown by the underline in FIG. 5, the text of [1], which is an example of the summary, is “soccer”, “XX representative”, “▽▽ stadium”, “□□□”, “representative”, “ This is a collection of word information of “World Cup”, “△△”, “Meeting”, “XX”, “Final Qualifying”, and “Challenge”.

  Then, as shown in [2], as the predetermined processing, here, the summaries included in the “seen” RSS feed are interested, while the summaries included in the “not seen” RSS feed are not interested. The user's interest in each word information is estimated from the word information included in the summary. When the user “sees” the sentence [1], the word information is included in the summary in which “soccer”, “Japan national team”, “▽▽”, “stadium”, etc. are considered interesting. On the other hand, it is assumed that word information such as “F”, “League”, “●● Cup”, “■■”, “◎◎◎” is included in the “not seen” summary example.

  Then, as a result of aggregating these word information for a plurality of summaries, as shown in FIG. 5, among the summaries with interest 133 and the uninterested summaries 2147, the interest summaries 133 include “ “Japan National Team” is included in 75, “ΔΔ” is included in 23, and “■■” is included in 4. On the other hand, the uninteresting summary 2147 includes 20 “Japan representatives”, 15 “ΔΔ”, and 279 “■■”. At this time, the word evaluation value for “Japan representative” is 1.7, the word evaluation value for “ΔΔ” is 1.3, and the word evaluation value for “■■” is −0.7. It is assumed that a learning (estimation) result is obtained based on processing by the learning unit 25 and the interest degree evaluation unit 27.

  Then, when the interest degree evaluation unit 27 acquires a summary including each word information of “F”, “League”, “Third act”, and “◆◆◆◆◆”, Evaluate the degree of interest for the word. Here, the degree-of-interest evaluation unit 27 determines that “F” is −0.1, “League” is 0.3, “Third Act” is 0.0, and “◆◆◆◆◆” is −1.4. The evaluation result of each word is obtained, and the overall summary is the evaluation result of -3.2.

  From the above, in the conventional method, the interest level evaluation process is also performed for the uninterested summary 2147. As a result, since a large amount of unread content that cannot be read by the user is learned, the amount of computation and the amount of information to be learned become enormous, and the load on the apparatus becomes excessive.

[Naive Bayesian transformation]
Therefore, the inventor of the present application newly found out that learning about unread content becomes unnecessary by converting the naive Bayes formula as described below. The conversion formula is shown below.

  The expression [1] is an expression for obtaining the likelihood λ of reading a sentence, and the probability P (read) of reading the summary in the total product of the appearance rates P (token | read) of word information in the read summary (content) Is divided by the sum of the appearance rate P (token | ignore) of the word information in the summary that has not been read and the probability P (ignore) that does not read the summary (denominator). Is required.

  Equation [2] is obtained by logarithmically transforming Equation [1]. The sum of P (token | read) logarithmically converted and the sum of P (read) logarithmically subtracted from the sum of P (token | ignore) logarithmically converted and the sum of P (ignore) Log (λ) is a summary evaluation value.

  Expression [3] is an arrangement of Expression [2]. log (λ) is the sum of P (token | read) logarithmically converted and P (token | ignore) subtracted from logarithm, and P (read) minus P (ignore). It will be the sum of In addition, since it is estimated that the thing which subtracted P (ignore) from P (read) becomes equal about all the summaries, when confirming the relative value about which of two summaries is seen, for example, You can ignore it.

  Equation [4] shows the contents of Σ in Equation [3]. Here, focusing on the word j, Tj indicates an evaluation value for the word j. Tj is a logarithm of the ratio of the read summary rj in which the word j appears with respect to the read summary number m (hereinafter referred to as “previous expression”), and the unread summary ij in which the word j with respect to the unread summary m ′ appears. Is obtained by subtracting the logarithm of the ratio (hereinafter referred to as “the following formula”). The “ratio of unread summaries in which word j appears with respect to summary m ′ that has not been read” in the following equation is the total appearance frequency of word j with respect to the sum of the number of summaries read. This is the ratio of rj minus the product with the total number of summaries.

  Equation [5] yields the sum of j from the previous equation of Equation [4] if M is sufficiently larger than m so that m is negligible for Equation [4], and therefore rj is negligible. It is possible to evaluate the degree of interest of the word j only by reducing the frequency. In other words, if the frequency of occurrence of each word can be obtained separately, the evaluation of the sentence can be approximated to the evaluation value in the equation [4] only by learning the read summary.

  Next, information stored in the appearance frequency table 14 of the management apparatus 10 will be described with reference to FIG. The information stored in the appearance frequency table 14 is a word character string 141, a word ID 142, a target RSS summary 143, and an appearance frequency 144, as shown in FIG.

  The word character string 141 is literally a character string of words and is information necessary for the content server 30. Note that the management device 10 and the information processing device 20 may be replaced with a word ID described later.

  The word ID 142 is a unique ID of a word assigned in the management apparatus 10. The word ID 142 may be replaced with a character string of the word. By converting the words into IDs, the amount of data can be compressed, and the record based on the learning information storage unit 26 in the information processing apparatus 20 can be set to a fixed length.

  The target RSS summary 143 is an RSS feed in which words appear. Although it is not necessary to store the database in the content server 30, it is indispensable when communicating between the management apparatus 10 and the information processing apparatus 20 or when caching in the information processing apparatus 20.

  The appearance frequency 144 is word appearance frequency information.

  Next, the interest degree evaluation procedure in the present embodiment will be described with reference to FIG. First, the content acquisition unit 11 acquires content from the content server 30 (step S1).

  Next, the feature amount totaling unit 13 calculates the appearance frequency of word information (step S2). The calculated appearance frequency is stored in the appearance frequency table 14 in the state shown in FIG.

  Thereafter, the word information and the appearance frequency are transmitted to the information processing apparatus 20 by the appearance frequency transmission unit 15 (step S3). In the information processing apparatus 20, the appearance frequency received by the appearance frequency receiving unit 22 is stored in the appearance frequency storage unit 24.

  On the other hand, the learning unit 25 records whether or not a predetermined process has been performed on the content received by the content receiving unit 21 (step S4). The learning unit 25 does not record (learn) content that has not been subjected to predetermined processing. However, if the amount of content that has undergone predetermined processing or the amount of content that has not been processed is small and, for example, sufficient learning information has not been obtained, learning is performed on content that has not been subjected to predetermined processing. Also good.

  The learning information storage unit 26 stores, as learning information, the probability that word information is included in the content that has undergone predetermined processing (step S5). Note that the word information for which the probability is already stored as learning information in the learning information storage unit 26 may be updated based on the probability that the content newly acquired by the learning unit 25 includes the same word information.

  The interest level evaluation unit 27 evaluates the interest level for the content acquired based on the appearance frequency stored in the appearance frequency storage unit 24 and the learning information stored in the learning information storage unit 26 (step S6).

[Modification]
Next, an information processing system according to a modification of the present embodiment will be described with reference to FIG. Note that the description of the same configuration as the information processing system in the present embodiment described above is omitted. The difference from the above-described embodiment is that the management apparatus 100 includes a learning unit 125, a learning information storage unit 126, an interest degree evaluation unit 127, and a user content management unit 128, and the information processing apparatus 200 includes the book described above. That is, the appearance frequency receiving unit 22, the appearance frequency storage unit 24, the learning unit 25, the learning information storage unit 26, and the interest degree evaluation unit 27 in the embodiment are not provided. The user content management unit 128 stores content information recommended for each user. In addition, although the function part enclosed with a broken line is collectively set as the calculating part 120 classified by user, it is only a formal name.

  In the present modification, the management device 100 only receives input data as to whether or not predetermined processing has been performed on the content via the input unit 28 of the information processing device 200, and the management device 10 performs learning processing and interest evaluation. By performing the processing, the processing load on the information processing apparatus 200 can be further reduced.

  Each of the above-described embodiments is a preferred embodiment of the present invention, and various modifications can be made without departing from the scope of the present invention. For example, each process in the information processing system of the present embodiment described above can be executed using hardware, software, or a combined configuration of both.

  In the case of executing processing using software, it is possible to install and execute a program in which a processing sequence is recorded in a memory in a computer incorporated in dedicated hardware. Alternatively, the program can be installed and executed on a general-purpose computer capable of executing various processes.

DESCRIPTION OF SYMBOLS 10,100 Management apparatus 11 Content acquisition part 12 Content management part 13 Feature-value totaling part 14 Appearance frequency table 15 Appearance frequency transmission part 16,116 Communication part 20,200 Information processing apparatus 21 Content reception part 22 Appearance frequency reception part 23 Content storage Unit 24 Appearance frequency storage unit 25 Learning unit 26, 126 Learning information storage unit 27, 127 Interest level evaluation unit 28 Input unit 29 Output unit 30 Content server 128 User content management unit

Claims (9)

An information processing system in which a management device that manages the appearance frequency of feature quantities included in acquired content and an information processing device that learns the user's interest in the content are connected via a network,
The management device
Appearance frequency calculating means for calculating the appearance frequency of the feature amount included in the acquired content;
Appearance frequency transmitting means for transmitting the feature amount included in the content and the appearance frequency calculated by the appearance frequency calculating means to the information processing apparatus,
The information processing apparatus includes:
Receiving means for receiving the feature amount included in the content and the appearance frequency calculated by the appearance frequency calculating means;
First storage means for storing the appearance frequency received by the receiving means in association with the feature amount included in the content;
Second storage means for storing, as learning information, a probability that a feature amount included in the content received by the receiving means is included in content that has undergone predetermined processing;
Learning means for updating learning information stored in the second storage means based on a feature amount included in the content for which the predetermined processing has been executed;
An interest degree evaluation means for evaluating an interest degree with respect to a feature amount included in the acquired content based on the appearance frequency and the learning information stored in the second storage means;
An information processing system comprising:   The degree-of-interest evaluation unit includes a probability that the feature amount is included in content that has been subjected to predetermined processing as learning information stored in the second storage unit, content that has been subjected to the predetermined processing, and predetermined content The information processing system according to claim 1, wherein the degree of interest in the feature amount included in the acquired content is evaluated based on a difference from the probability that the feature amount is included in content that has not been executed. . The management device includes a conversion unit that converts a feature amount included in the acquired content into predetermined identification information,
The appearance frequency transmitting means includes the identification information converted from the feature amount included in the content by the conversion means, and the appearance frequency of the feature amount included in the content calculated by the appearance frequency calculation means. Sent to the processing unit,
The receiving unit receives the identification information converted from the feature amount included in the content and the appearance frequency calculated by the appearance frequency calculating unit;
3. The information processing according to claim 1, wherein the first storage unit stores the appearance frequency received by the receiving unit in association with identification information converted from a feature amount included in the content. 4. system.   4. The information processing system according to claim 1, wherein the appearance frequency transmitting unit transmits a predetermined feature amount as a feature amount included in the content to the information processing apparatus. 5. . An information processing system in which a management device that manages the appearance frequency of feature quantities included in acquired content and an information processing device that learns the user's interest in the content are connected via a network,
Appearance frequency calculating means for calculating the appearance frequency of the feature amount included in the acquired content;
First storage means for storing the appearance frequency calculated by the appearance frequency calculation means in association with the feature amount included in the content;
Second storage means for storing, as learning information, a probability that a feature amount included in the content stored in the first storage means is included in the content that has undergone predetermined processing;
Learning means for updating learning information stored in the second storage means based on a feature amount included in the content for which the predetermined processing has been executed;
An interest degree evaluation means for evaluating an interest degree with respect to a feature amount included in the acquired content based on the appearance frequency and the learning information stored in the second storage means;
An information processing system comprising: An information processing method of an information processing system in which a management device that manages the appearance frequency of a feature amount included in acquired content and an information processing device that learns a user's interest in the content are connected via a network,
The management device
Calculating the appearance frequency of the feature amount included in the acquired content;
Run and sending the occurrence frequency calculated feature amount included in the content to the information processing apparatus,
The information processing apparatus includes:
Receiving the feature amount included in the content and the calculated appearance frequency;
Storing the received appearance frequency in the first storage unit in association with the feature amount included in the content;
Storing the probability that the feature amount included in the received content is included in the content that has been subjected to the predetermined processing in the second storage unit as learning information;
Updating the learning information stored in the second storage unit based on the feature amount included in the content for which the predetermined processing has been performed;
Evaluating the degree of interest in the feature amount included in the acquired content based on the appearance frequency and the learning information stored in the second storage unit;
The information processing method characterized by performing . An information processing method of an information processing system in which a management device that manages the appearance frequency of a feature amount included in acquired content and an information processing device that learns a user's interest in the content are connected via a network,
The information processing system includes:
Calculating the appearance frequency of the feature amount included in the acquired content;
Storing the calculated appearance frequency in the first storage unit in association with the feature amount included in the content;
Storing in the second storage unit the probability that the feature amount included in the content stored in the first storage unit is included in the content that has undergone predetermined processing as learning information;
Updating the learning information stored in the second storage unit based on the feature amount included in the content for which the predetermined processing has been performed;
Evaluating the degree of interest in the feature amount included in the acquired content based on the appearance frequency and the learning information stored in the second storage unit;
The information processing method characterized by performing . A computer-readable program that is executed by an information processing system in which a management device that manages the appearance frequency of a feature amount included in acquired content and an information processing device that learns the user's interest in the content are connected via a network Because
The management device
A process of calculating the appearance frequency of the feature amount included in the acquired content;
Processing to transmit the feature amount included in the content and the calculated appearance frequency to the information processing apparatus,
The information processing apparatus includes:
Processing for receiving the feature amount included in the content and the calculated appearance frequency;
A process of storing the received appearance frequency in the first storage unit in association with the feature amount included in the content;
A process of storing the probability that the feature amount included in the received content is included in the content for which the predetermined process has been executed in the second storage unit as learning information;
A process of updating learning information stored in the second storage unit based on a feature amount included in the content for which the predetermined process has been executed;
Based on the appearance frequency and the learning information stored in the second storage unit, a process for evaluating the degree of interest in the feature amount included in the acquired content;
The program characterized by including. A computer-readable program that is executed by an information processing system in which a management device that manages the appearance frequency of a feature amount included in acquired content and an information processing device that learns the user's interest in the content are connected via a network Because
A process of calculating the appearance frequency of the feature amount included in the acquired content;
A process for storing the calculated appearance frequency in the first storage unit in association with the feature amount included in the content;
A process of storing, in the second storage unit, the probability that the feature amount included in the content stored in the first storage unit is included in the content that has been subjected to a predetermined process as learning information;
A process of updating learning information stored in the second storage unit based on a feature amount included in the content for which the predetermined process has been executed;
Based on the appearance frequency and the learning information stored in the second storage unit, a process for evaluating the degree of interest in the feature amount included in the acquired content;
The program characterized by including.

Images