JP2018136931A - Distribution system, distribution method and distribution program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a distribution system capable of evaluating influences on user's impression to a given object by delivered information.SOLUTION: Disclosed distribution system includes: a generation part that generates a distributed expression for similar contexts which is a distributed expression of the context reflected with impression of users represented by the text, and which relatively represents impression of the users based on a content which includes a text representing an impression of a user to the context; and a providing part that provides a piece of information representing changes in relationship among the distributed expression of the given context, which is generated before a predetermined event relevant to a given context occurs, the distributed expression of plural contexts as reference, a distributed expression of the given context which is generated after the predetermined event occurs and the distributed expression of plural contexts as reference.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a providing device, a providing method, and a providing program.

  In recent years, with the rapid spread of the Internet, information distribution via a network has been actively performed. As an example of such information distribution, a technique for distributing information on a predetermined target such as an advertisement or news (hereinafter referred to as “distribution information”) is known.

  Also, a method for evaluating the influence of distribution information on users has been proposed. For example, there is known a method for evaluating how much the distribution information conveys information related to a predetermined target to the user based on the number of times the distribution information is browsed, the number of times the distribution information is selected by the user, and the like.

JP, 2006-207141, A

  However, it is difficult to say that the above-described conventional technology appropriately evaluates the influence of distribution information on the user's impression of a predetermined target.

  In other words, the conventional technology only evaluates how much the distribution information conveys to the user information related to the predetermined target, and evaluates how the user's impression of the predetermined target has been changed by the distribution information. It may not be said that you are doing.

  The present application has been made in view of the above, and an object thereof is to provide information for evaluating the influence of distribution information on a user's impression of a predetermined target.

  The providing apparatus according to the present application is a distributed representation of the context reflecting the user's impression indicated by the text based on content including the text indicating the user's impression of the context, and the user's impression A generation unit that generates a distributed expression for each context of the same type, a distributed expression of the predetermined context generated before a predetermined event related to the predetermined context occurs, and a distributed expression of a plurality of reference contexts Provides information indicating a change in the relationship between the relationship between the distributed representation of the predetermined context generated after the predetermined event and the distributed representation of the plurality of reference contexts And a providing unit.

  According to an aspect of the embodiment, it is possible to evaluate the influence of the distribution information on the user's impression of the predetermined target.

FIG. 1 is a diagram illustrating an example of a providing process executed by the information providing apparatus according to the embodiment. FIG. 2 is a diagram illustrating a configuration example of the information providing apparatus according to the embodiment. FIG. 3 is a diagram illustrating an example of information registered in the log database according to the embodiment. FIG. 4 is a diagram illustrating an example of information registered in the distributed representation space database according to the embodiment. FIG. 5 is a diagram illustrating an example of information generated by the information providing apparatus according to the embodiment. FIG. 6 is a diagram illustrating an example of information indicating a history of changes in distributed representation provided by the information providing apparatus according to the embodiment. FIG. 7 is a flowchart illustrating an example of a flow of providing processing executed by the information providing apparatus according to the embodiment. FIG. 8 is a diagram illustrating an example of a hardware configuration.

  Hereinafter, modes for implementing a providing device, a providing method, and a providing program according to the present application (hereinafter referred to as “embodiments”) will be described in detail with reference to the drawings. In addition, the providing apparatus, the providing method, and the providing program according to the present application are not limited by this embodiment. In the following embodiments, the same parts are denoted by the same reference numerals, and redundant description is omitted.

[Embodiment]
[1. Regarding the processing provided by the information providing device]
First, an example of a providing process executed by an information providing apparatus as an example of a providing apparatus will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of a providing process executed by the information providing apparatus according to the embodiment.

  In the following description, as the providing process executed by the information providing apparatus 10, in order to evaluate how the user's impression on the predetermined target has changed due to distribution information such as advertisements and news regarding the predetermined target. An example of a process for providing the information will be described. Here, the predetermined target is any product, service, company, store, brand, facility, etc., as long as the information is transmitted to any user by distribution information such as advertisements and news, Any object can be adopted. The predetermined target may be, for example, a person such as an actor, various contents such as animals, movies and music. In the following description, a predetermined target related to distribution information is referred to as a “determination target”. Further, in the following description, content related to the determination target, such as various contents such as text, images, and sounds that remind the user of various determination targets, is described as a context indicating the determination target.

  In addition to the advertisement and the news, any information can be adopted as the distribution information as long as the information related to the determination target is transmitted to any user. The advertisement is a concept including not only a commercial or non-profit advertisement but also volunteer recruitment, public advertisement, public notice, and other arbitrary contents. In addition to content that includes so-called advertisement-related information, distribution information is included in information related to an arbitrary target or content related to an arbitrary target (such as a landing page) that interests the user. As long as the information is widely notified, images, moving images, text, graphics, symbols, hyperlinks, and other arbitrary contents may be included together with the text. In the following description, an example of a providing process executed when an advertisement related to a predetermined brand is distributed as distribution information will be described as distribution information.

[1-1. Example of information providing device)
The information providing apparatus 10 can communicate with the information management server 100, the distribution server 200, the consumer server 300, and the user terminal 400 via a predetermined network N (for example, see FIG. 2) such as the Internet. For example, the processing device is realized by a server device, a cloud system, or the like. The information providing apparatus 10 may be able to communicate with an arbitrary number of information management servers 100, distribution servers 200, consumer servers 300, and user terminals 400 via the network N.

  The information management server 100 is an information processing device that provides an arbitrary web service to a user, and is realized by, for example, a server device or a cloud system. For example, the information management server 100 is a server device that provides various social networking services (SNSs) that provide information posted by a user to other users. When the information management server 100 accepts postings of various contents such as blogs, microblogs, web pages, messages, still images, moving images, sounds, etc., the information management server 100 publishes the accepted contents, thereby providing various information related to the contents. It may be a server device that spreads to the general public.

  The information management server 100 may be a server device that provides information about various dictionaries, news, and the like to the user. In addition, the information management server 100 is an arbitrary web service such as a portal site, Internet auction, online shopping mall, web search, route search, map search, game, real estate information provision, financial information provision, accommodation facility reservation service, etc. It may be a server device that provides In addition, the information management server 100 may be a server device that accepts a user's evaluation for an arbitrary target and provides a service that provides an evaluation result. Further, the information management server 100 may be a server device that collects a log indicating an action history in the real world or a network such as a user's location information and purchase history. Such an information management server 100 manages various contents posted by an arbitrary user, an action history of an arbitrary user, a search history of an arbitrary user, information on an electronic commerce target, and the like. .

  The distribution server 200 is a server device that distributes various types of distribution information such as advertisements and news to the user terminal 400, and is realized by, for example, a server device or a cloud system. For example, when distributing information related to advertisements, the distribution server 200 lists advertisements to be distributed based on demographic attributes, psychographic attributes, and the like of the user who uses the user terminal 400, and displays the listing results. The corresponding advertisement is distributed to the user terminal 400. In addition, as long as delivery information is delivered to the user terminal 400, the delivery server 200 may deliver not only advertisements but also arbitrary information. Further, the distribution server 200 may be realized as one unit with the information management server 100.

  The consumer server 300 is an information processing apparatus used by a consumer who desires information indicating the influence of distribution information, and is realized by a server apparatus, a cloud system, or the like. For example, when an advertisement related to a predetermined target is distributed as distribution information, the consumer server 300 is used by a consumer who desires the influence of an advertisement such as an advertiser on the user.

  The user terminal 400 is a terminal device used by an arbitrary user, and is realized by an information processing device such as a PC (Personal Computer), a smart device, a mobile terminal device, a server device, a cloud system, or the like. For example, the user terminal 400 displays the content of the distribution information by displaying a function for transmitting a post entered by the user to the information management server 100, a function for displaying information managed by the information management server 100, and distribution information. Has a function to transmit the message to the user.

[1-2. Example of provision processing)
Here, a method for evaluating the influence of various distribution information such as advertisements and news on users has been proposed. For example, a method is known that evaluates how much information related to an object related to the distribution information is transmitted to the user based on the number of times the distribution information is browsed, the number of times the distribution information is selected by the user, and the like. However, such conventional technology only evaluates how much the distribution information conveys information about the determination target to the user, and how the impression of the user with respect to the determination target is changed by the distribution information. It may not be said that it is evaluating.

  For example, it may be desired to distribute an advertisement that induces a brand image that a company has (that is, a user's impression of a determination target). When such an advertisement is delivered, the prior art only shows how much the advertisement has been viewed, etc., whether the advertisement actually changed the brand image, that is, the impression of the user to the company Is not an appropriate indication of whether or not the advertisement has actually changed.

  Therefore, the information providing apparatus 10 executes the following providing process. First, the information providing apparatus 10 generates a distributed representation of each context based on the relative relevance of a plurality of contexts. Then, the information providing apparatus 10 includes a distributed representation of the predetermined context generated before distribution information related to the predetermined context is distributed, and a distributed representation of the predetermined context generated after the distribution information is distributed. Provide information indicating changes in

  For example, the information providing apparatus 10 acquires various types of information managed by the information management server 100 and indicates various determination targets related to distribution information, such as company name, product name, service name, brand name, facility name, and person name. Various contents such as text and images are defined as contexts indicating determination targets (hereinafter referred to as “determination target contexts”). And the information provision apparatus 10 produces | generates the distributed expression of each context using the technique which converts each context into a multidimensional quantity based on the relative relevance which each context has, such as w2v (word2vec). Note that the information providing apparatus 10 may adopt any technique other than w2v as long as it is a technique that converts each context into a multidimensional quantity based on the relative relevance of each context.

  Further, for example, the information providing apparatus 10 receives from the consumer server 300 the determination target context and the date and time when the distribution information that advertises the determination target is distributed. In such a case, the information providing apparatus 10 is a distributed expression corresponding to the determination target context, and is a distributed expression generated from information before distribution information is distributed and a distributed expression corresponding to the determination target context. And the distributed representation generated from the information after the distribution information is distributed.

  Here, when the impression of the user with respect to the determination target is changed by the distribution information, the distributed expression corresponding to the determination target context is considered to be a different distributed expression before and after the distribution information is distributed. More specifically, when the distributed representation of an arbitrary context is selected as a reference, the distributed representation of the determination target context changes before and after the distribution information is distributed with respect to the distributed representation of the reference context. It is thought that.

  For example, it is assumed that brand A and brand B exist, and distribution information that brings the image of brand B closer to brand A is distributed. In such a case, if the distribution information brings the image of brand B closer to brand A, the distributed representation of brand B will change to be similar to the distributed representation of brand A by distributing the distribution information. . Also, if the distribution information does not bring the image of brand B close to or away from brand A, the distributed representation of brand B changes so that it does not resemble the distributed representation of brand A by distributing the distribution information. It is thought that.

  Therefore, the information providing apparatus 10 determines that information indicating a change in the relevance between the distributed representation of the reference context (hereinafter referred to as “reference context”) and the distributed representation of the determination target context is the distribution target. Is output as an index indicating whether or not the user's impression is changed. As a result, the information providing apparatus 10 can provide information indicating whether or not the distribution information has changed the user's impression.

[1-3. About generation of distributed representation)
Here, processing for generating a distributed representation of each context will be described. For example, the information providing apparatus 10 may be any arbitrary information such as content posted by a user, search query history, news, dictionary, user behavior information, etc., from an arbitrary server apparatus such as the information management server 100. To collect. In the following description, various types of information collected by the information providing apparatus 10 to generate a distributed representation may be described as “log”. Subsequently, the information providing apparatus 10 extracts an arbitrary context from the log by executing processing such as morphological analysis. Note that the context is not limited to the context indicating the determination target.

  Subsequently, the information providing apparatus 10 distributes from the context so that the distributed expressions of similar contexts are similar and the distributed expressions of dissimilar contexts are not similar based on the appearance frequency of other words appearing with each context. Learn a learner that generates an expression. That is, the information providing apparatus 10 generates a learning device that performs vectorization (that is, distributed expression) of each context based on the relative relevance of a plurality of contexts. And the information provision apparatus 10 produces | generates the distributed expression of each context using the produced | generated learning device. In the following description, a set of distributed representations of each context may be described as a distributed representation space.

  Here, the information posted by an arbitrary user may include a word indicating the user's impression of the determination target, such as “A is cute” or “B is high-class”. When the distributed expression space of each context is generated from such information, it is considered that the user's impression of the determination target is reflected in the distributed expression of each determination target context. More specifically, it is considered that distributed expressions similar to each other are generated from corresponding contexts for a plurality of determination targets for which the user has a high-quality impression.

  Therefore, the information providing apparatus 10 generates a distributed representation of each context from information posted by the user using a technology that converts each context into a multidimensional quantity based on the relative relevance of each context. . As a result, the information providing apparatus 10 can generate a distributed representation of each context based on the relative relevance of an impression that an arbitrary user has with respect to a determination target corresponding to each context.

  In addition, the information provision apparatus 10 is a user's impression with respect to each determination target, such as content posted by an arbitrary user, an arbitrary user's action history, an arbitrary user's search history, or information on an electronic commerce target. As long as the information can be included, the distributed representation space of each context may be generated based on arbitrary information.

  Further, the information providing apparatus 10 need not reflect the impression of the user with respect to the determination target based on the same post, an adjective included in the same sentence, and the like. That is, the information providing apparatus 10 not only includes information posted by the user U, but also, for example, the impression of the user U with respect to a determination target regarding posting based on biometric information and action content when the user U posted. And a distributed expression reflecting the estimated impression may be generated. Also, based on whether or not the user U frequently visits a store or a place related to the determination target, the user U's impression on the determination target is estimated, and a distributed expression reflecting the estimated impression is generated. May be. That is, the information providing apparatus 10 may generate a distributed expression from arbitrary information by an arbitrary method as long as the relative impression of the user U with respect to the determination target can be reflected in the distributed expression.

[1-4. (Example of provision processing)
Then, an example of the provision process which the information provision apparatus 10 performs is demonstrated. In the following example, an example of processing for providing information for evaluating whether or not the distribution information has changed the impression when the distribution information regarding the brand “A” is distributed will be described. In the following description, “user U” does not indicate a specific user, but indicates an unspecified number of users.

  For example, the user U posts various information related to the brand “A” to the information management server 100 (step S1). For example, the user U posts information including an impression about A such as “A is cute”. In such a case, the information providing apparatus 10 acquires various posts (step S2). And the information provision apparatus 10 produces | generates the dispersion | distribution expression based on the relative relevance of the impression which a user has with respect to each brand before distribution information, such as an advertisement and news, is distributed from the acquired various postings. (Step S3).

  For example, the information providing apparatus 10 performs analysis such as morphological analysis from various information posted by the user U such as “A is cute”, “B is high-class”, “A is a popular product”, “C is rare”, etc. Using technology, “A”, “B”, and “C”, which are contexts indicating various brands, are extracted. Then, the information providing apparatus 10 uses w2v to convert “A”, “B”, and “C” into distributed expressions so as to reflect the relative relevance of the impression that the user has. For example, the information providing apparatus 10 uses various learning devices “A”, “B”, “C” using a learning device that converts the context of a brand into a distributed expression so that the distributed expressions of brands having similar impressions are similar. Is generated. Then, the information providing apparatus 10 generates a distributed expression space VS1 including a distributed expression of the context of each brand (hereinafter referred to as “distributed expression of each brand”). In the example illustrated in FIG. 1, an example of the distributed expression space including the distributed expressions of the brands “A” to “E” is described.

  Next, the distribution server 200 distributes distribution information such as advertisements and news of the brand “A” to the user terminal 400 of the user U (step S4). Further, the information management server 100 accepts again various posts posted by the user U after distribution information distribution (step S5). Further, the information providing apparatus 10 obtains various new posts from the user U (step S6). And the information provision apparatus 10 produces | generates the distributed expression based on the relative relevance of the impression which a user has about each brand (namely, context which shows each brand) after distribution information is distributed (step S7). ). For example, the information providing apparatus 10 determines each of the various information posted by the user U such as “A is good”, “B is high-quality”, “A is good quality”, “A is a high-class route?” A distributed representation of the brand is generated, and a distributed representation space VS2 including the generated distributed representation is generated.

  Here, since it is considered that distribution information related to various brands is distributed to the user U, the user's impression of each brand may change with the passage of time. As a result, the distributed representation of each brand is considered to change every time the distributed representation is generated, so it may be difficult to evaluate distribution information using only a single brand's distributed representation. . However, for example, if the distribution information is information that can bring the impression of brand “A” closer to the impression of brand “B”, at least after the distribution information is distributed as shown in FIG. The distributed representation of brand “A” is closer to the distributed representation of brand “B” than the distributed representation of brand “A” before distribution of distribution information.

  Therefore, the information providing apparatus 10 generates content indicating how the classification information of the determination target context has changed with respect to a predetermined reference context as information for evaluating distribution information (step S8). For example, the information providing apparatus 10 determines the difference between the distributed representation of the reference context generated before the distribution information is distributed and the distributed representation of the determination target context, and the reference context generated after the distribution information is distributed. Information indicating a change between the distributed representation and the difference between the distributed representation of the determination target context is generated. More specifically, the information providing apparatus 10 generates information indicating a change in relevance between a distributed representation of a plurality of reference contexts and a distributed representation of a predetermined context.

  For example, the information providing apparatus 10 selects the context of the brand “A” related to the distribution information as the determination target context. Further, the information providing apparatus 10 selects the contexts of the brands “B”, “C”, and “E” as the reference context. The selection target context and the reference context may be selected from a consumer who uses the consumer server 300.

  Then, the information providing apparatus 10 uses the first normalization function for projecting the distributed representation of the reference context before the distribution information is distributed to a predetermined position, and the distributed representation of the reference context after the distribution information is distributed. A second normalization function that projects to a predetermined position is generated. Then, the information providing apparatus 10 determines the position where the distributed representation of the determination target context before the distribution information is distributed is projected by the first function and the distributed representation of the determination target context after the distribution information is distributed. Information indicating the position projected by the function of 2 is provided. That is, the information providing apparatus 10 identifies the change in the distributed representation of the determination target context by normalizing the change in the distributed representation of the reference context using the normalization function.

For example, in the distributed representation space VS1 generated from post information before distribution information is distributed, the information providing apparatus 10 distributes the reference context, that is, the contexts of brands “B”, “C”, and “E”. Each of the distributed expressions is specified. Then, the information providing apparatus 10 calculates a normalization function F ₁ (x) for projecting the distributed representation of the reference context in the distributed representation space VS1 to each predetermined position. For example, the information providing apparatus 10 converts the distributed representations of the brands “B”, “C”, and “E” in the distributed representation space VS1 into equilateral triangles on a predetermined two-dimensional plane (hereinafter referred to as “output plane”). A normalization function F ₁ (x) to be projected onto each vertex is calculated. Then, the information providing apparatus 10 projects the distributed representation of the brand “A” in the distributed representation space VS1 on the output plane using the normalization function F ₁ (x). As a result, the information providing apparatus 10 generates a graph indicating the impression of the brand “A” that the user has relative to the brands “B”, “C”, and “E” before the distribution information is distributed. Can do.

Subsequently, the information providing apparatus 10 uses the distributed representation of the reference context, that is, the brands “B”, “C”, and “E” in the distributed representation space VS2 generated from the post information after the distribution information is distributed. Identify each distributed representation of the context. Then, the information providing apparatus 10 normalizes the function F ₂ (x) that projects the distributed representations of the brands “B”, “C”, and “E” in the distributed representation space VS2 onto the vertices of the regular triangle on the output plane. Is calculated. Then, the information providing apparatus 10 projects the distributed representation of the brand “A” in the distributed representation space VS2 on the output plane using the normalization function F ₂ (x). As a result, the information providing apparatus 10 generates a graph indicating the impression of the brand “A” that the user has relative to the brands “B”, “C”, and “E” after the distribution information is distributed. be able to.

  And the information provision apparatus 10 produces | generates the information which shows the change of the position where a determination target context is projected, when the dispersion | distribution expression of several reference | standard context is each projected on the predetermined position. For example, the information providing apparatus 10 generates a graph C10 in which graphs generated from the distributed representation space VS1 and the distributed representation space VS2 are superimposed so that the brands corresponding to the vertices match. That is, the information providing apparatus 10 identifies a change in the relative relationship between the reference context and the determination target context by normalizing the relative relationship of the plurality of contexts for each distributed expression space.

  Then, the information providing apparatus 10 has an arrow from the position where the distributed representation of the brand “A” before distribution of distribution information is projected to the position where the distributed expression of brand “A” after distribution of distribution information is projected. Is set on the graph C10. As a result, the information providing apparatus 10 changes the impression of the brand “A” that the user has relative to the brands “B”, “C”, and “E” as shown in FIG. Can be generated.

  Thereafter, the information providing apparatus 10 distributes the generated content to the consumer server 300 as content for evaluating the influence of the distribution information on the user's impression (step S9). As a result, the consumer can determine whether the distribution information distributed in step S4 actually changes the impression of the brand “A”.

  For example, if the distributed representation of the brand “A” is approaching the distributed representation of the brand “B”, the consumer information about the distribution of the brand “A” will change the impression of the brand “A” to the impression of the brand “B”. It can be determined that it has approached. Also, for example, if the distributed representation of the brand “A” is far from the distributed representation of the brand “E”, the distribution information of the brand “A” gives the impression of the brand “A” to the brand “E”. It can be determined that it has moved away from the impression. In addition, when the distributed representation of the brand “A” is not moved with respect to the brands “B”, “C”, and “E”, the consumer can send the distribution information of the brand “A” to the brand “B”. It can be determined that the impression of the brand “A” was not changed based on the impressions of “C” and “E”. That is, the consumer can determine whether or not the distribution information has changed the user's impression.

  Note that the information provided by the information providing apparatus 10 may be used for evaluating distribution information in an arbitrary manner. For example, if the sales of the brand “A” rises even though the distributed representation of the brand “A” has not changed much before and after distribution of the distribution information, the distribution information changes the brand image. It may be determined that sales are increasing without any problems. In addition, if the distributed representation of the brand “A” is close to the distributed representation of the brand “B” after the distribution information is distributed, the consumer can change the brand image of the brand “A” to “brand B”. You may determine that it is approaching.

  As described above, the information providing apparatus 10 generates a distributed representation of each context based on the relative relevance of a plurality of contexts. Then, the information providing apparatus 10 includes a distributed representation of the predetermined context generated before the distribution information regarding the predetermined context is distributed, and a distributed representation of the predetermined context generated after the distribution information is distributed. Provide information indicating changes. Therefore, the information providing apparatus 10 can provide a guideline for evaluating the user's impression of a predetermined context such as the context of the brand “A”.

[2. (Variation of provision processing)
In the above, an example of the providing process by the information providing apparatus 10 has been described. However, the embodiment is not limited to this. Hereinafter, the variation of the provision process which the information provision apparatus 10 performs is demonstrated.

[2-1. (About user impressions)
In the example described above, the information providing apparatus 10 converts the determination target context into the distributed expression, thereby reflecting the impression of the user with respect to the determination target in the distributed expression. This process is a process for reflecting an impression on the determination target itself, and also a process for reflecting an impression on the determination target context. That is, the user impression reflected in the distributed expression is a concept that includes not only the impression of the determination target itself but also the impression of the context indicating the determination target.

  For example, when there are a plurality of texts having the same brand name and different fonts, the information providing apparatus 10 may treat these texts as individual contexts or treat them as the same context. Good. By executing such processing, the information providing apparatus 10 may reflect the impression of the brand itself (that is, the determination target) in the distributed expression, or reflect the impression of each context indicating the brand in the distributed expression. May be.

[2-2. (Information to be provided)
In the example described above, the information providing apparatus 10 provides information indicating a change in the user's impression on the determination target. Here, each time the distribution information is distributed, the information providing apparatus 10 generates a distributed representation space, and determines the positional relationship between the distributed representation of the reference context and the distributed representation of the determination target context from each of the generated distributed representation spaces. Information indicating the change in the user's impression in time series may be generated by generating the information indicating. That is, when the distribution information is distributed a plurality of times, the information providing apparatus 10 may provide information indicating a change in the distributed representation of the determination target context and occurring every time the distribution information is distributed. . For example, the information providing apparatus 10 may generate and provide information indicating a history of changes in the position at which the determination target context is projected, that is, a history of changes in the impression of the user with respect to the determination target.

  The information providing apparatus 10 may generate information indicating the positional relationship between the distributed representation of the reference context and the distributed representation of the determination target context each time the distribution information is distributed, and the timing at which the distribution information is distributed. Regardless of, information indicating the positional relationship may be generated. For example, the information providing apparatus 10 generates a distributed representation space at a predetermined time interval, and generates information indicating the positional relationship between the distributed representation of the reference context and the distributed representation of the determination target context from each of the generated distributed representation spaces. . And the information provision apparatus 10 may show a consumer how the user's impression with respect to the determination object changed by outputting the information which shows the transition of the produced | generated positional relationship.

  Further, the information providing apparatus 10 may use an arbitrary number of contexts as reference contexts. For example, when n (natural number) number is selected as the reference context, the information providing apparatus 10 generates a normalization function that projects each reference context on each vertex of the n-gon, and uses the generated normalization function. By projecting a distributed representation of the determination target context, information indicating a change in impression that the user has relative to each reference context and the determination target context may be generated. Further, the information providing apparatus 10 may automatically select the reference context so that the distributed representation of the determination target context is projected on the n-gonal inner region. Note that when n reference contexts are selected, projection may be performed on an n-1 dimensional simplex. For example, when projecting from an n-1 dimensional simplex to an n-gon, the simplex A normalization function for performing projection using a predetermined projection method (that is, an ad hoc projection method) set for each combination of the number of dimensions and the number of vertices of a polygon as a projection destination may be generated. For such a projection method, any method can be applied.

  In addition, when there are two reference contexts, the information providing apparatus 10 projects the distributed representation of the determination target context on the line segment where each reference context is projected, so that the impression of the determination target is Information indicating which side of the reference context is used may be generated.

  Further, the information providing apparatus 10 may use a context belonging to a category common to the determination target context as a reference context. For example, a food company and a technology-related company are considered to have different relevance of relative impressions of users. Therefore, for example, when the determination target context is a context related to a food company, the information providing apparatus 10 may select a context related to another food company as the reference context.

  In addition, when the contexts of food companies and technology-related companies are combined and converted into a distributed representation, there is a risk that an error will occur between the impression that the user actually has for each company and the impression reflected in the distributed representation. is there. Therefore, when the determination target context is a food company, the information providing apparatus 10 may generate a distributed expression space in which the context of each food company is a distributed expression.

  Further, the information providing apparatus 10 may output the information indicating the distributed representation space VS1 and the information indicating the distributed representation space VS2 as they are. For example, the information providing apparatus 10 may project each of the distributed representation spaces VS1 and VS2 onto a two-dimensional plane and generate and provide information indicating a position where the distributed representation of the determination target context is projected. The information providing apparatus 10 may exclude information indicating a distributed representation of contexts other than the reference context and the determination target context from the plurality of contexts from the provision target. For example, the information providing apparatus 10 projects each of the distributed expression spaces VS1 and VS2 onto a two-dimensional plane, and excludes the distributed expressions other than the distributed expression of the determination target context and the distributed expression of the reference context from the display target. Also good.

  Further, the information providing apparatus 10 may evaluate the distribution information based on the change in the distributed representation of the determination target context, and provide the evaluation result of the distribution information as information indicating the change in the distributed representation. For example, when the distribution information for bringing the impression of the brand “A” close to the impression of the brand “B” is distributed, the information providing apparatus 10 distributes the distribution expression of the brand “A” and the brand before the distribution information is distributed. The distance between the distributed representation of “B” (eg, cosine distance and Hamming distance) and the distance between the distributed representation of brand “A” and the distributed representation of brand “B” after distribution information is distributed. . Then, when the distance after distribution is shorter than the distance before distribution, the information providing apparatus 10 indicates that the distribution information can bring the impression of the brand “A” closer to the impression of the brand “B”. If the distance after the distribution is longer than the distance before the distribution, the distribution information indicates that the impression of the brand “A” could not approach the impression of the brand “B”. The evaluation result shown may be generated.

[2-3. About generation of distributed expression space)
Here, the information providing apparatus 10 may execute an arbitrary process when generating the distributed expression space in order to clarify the relationship between the distributed expression of the reference context and the distributed expression of the determination target context. For example, the information providing apparatus 10 may generate a distributed representation of each context so that the distributed representations of the reference context are separated from each other by a predetermined threshold or more. More specifically, when the reference context is selected in advance, the information providing apparatus 10 performs learning of a learning device that converts the context into the distributed representation so that the distributed representation of the reference context is not similar. Also good.

  Further, the information providing apparatus 10 may convert a context (hereinafter referred to as “anchor”), which is a guide (anchor) on the distributed expression space, into a distributed expression. For example, the information providing apparatus 10 extracts a predetermined adjective such as “high” or “cute” from a posting or the like as an anchor. Then, the information providing apparatus 10 generates a distributed expression of the determination target context and the anchor.

  When such a distributed expression space is generated, the anchor serves as one index of what impression the user has with respect to the determination target. Therefore, the information providing apparatus 10 may facilitate selection of the reference context by providing information indicating the distributed expression space including the anchor to the consumer. Further, the information providing apparatus 10 may provide information indicating a change in relationship between the distributed expression of a predetermined adjective and the distributed expression of the determination target context.

[2-4. About application mode]
Here, if the information providing apparatus 10 provides information indicating how the user's impression on a certain determination target has changed when distributing certain information such as distribution information, the above-described embodiment. Besides, it can be applied to any form.

  For example, the information providing apparatus 10 includes a distributed representation of the determination target context generated before the advertisement information for advertising the determination target context is distributed, and a distributed representation of the determination context generated after the advertisement information is distributed. Information indicating changes in the number may be provided. When such processing is executed, the information providing apparatus 10 can provide information indicating whether or not the advertisement can appropriately guide the user's impression of the determination target.

  Further, for example, the information providing apparatus 10 identifies a change in the distributed expression of a predetermined determination target context before and after the news is distributed as a distribution target, so that the news is a user for the predetermined determination target. Information indicating whether or not the impression has been changed may be provided. When such processing is executed, the delivered news and the determination target context may be directly related or may be indirectly related.

  Further, for example, the information providing apparatus 10 generates a distributed representation of the determination target context at a predetermined time interval, and determines whether or not the generated distributed representation has changed. Then, when the distributed representation has changed in a certain period, the information providing apparatus 10 may provide the consumer that the distribution information distributed in the period has changed the user's impression.

[3. Configuration of information providing device]
Hereinafter, an example of the functional configuration of the information providing apparatus 10 will be described. FIG. 2 is a diagram illustrating a configuration example of the information providing apparatus according to the embodiment. As illustrated in FIG. 2, the information providing apparatus 10 includes a communication unit 20, a storage unit 30, and a control unit 40.

  The communication unit 20 is realized by, for example, a NIC (Network Interface Card). The communication unit 20 is connected to the network N in a wired or wireless manner, and transmits / receives information to / from the information management server 100, the distribution server 200, the consumer server 300, and the user terminal 400.

  The storage unit 30 is realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. The storage unit 30 also stores a log database 31, a distributed representation space database 32, and a model database 33.

  In the log database 31, various posts used when generating a distributed expression are registered. For example, FIG. 3 is a diagram illustrating an example of information registered in the log database according to the embodiment. In the example illustrated in FIG. 3, information such as “log ID (Identifier)”, “log type”, “user ID”, “date / time”, “log contents”, and the like is registered in the log database 31.

  Here, the “log ID” is an identifier for identifying a log. The “log” is information indicating whether the log indicated by the associated “log ID” is a post such as SNS or a search query. The “user ID” is an identifier of a user who is related to the log indicated by the associated “log ID”. “Date and time” indicates the date and time when the log indicated by the associated “log ID” is posted on the network. The “log contents” are various contents collected as logs.

  For example, in the example illustrated in FIG. 3, the log database 31 includes a log ID “log ID # 1”, a log type “post”, a user ID “user # 1”, a date and time “date and time # 1”, and a log content “ Information such as “A is cute” is associated and registered. Such information is recorded as a log indicated by the log ID “log ID # 1” as content posted by the user indicated by the user ID “user # 1” on the date indicated by the date “date # 1”. The content “A is cute” is registered.

  In the example illustrated in FIG. 3, conceptual values such as “log ID # 1”, “user # 1”, and “date and time # 1” are described. And a character string for identifying the user and a character string indicating the date and time are registered. In addition to the information shown in FIG. 3, arbitrary information may be registered in the log database 31 as long as it is information related to various types of information collected as a log. Further, in the example shown in FIG. 3, an example in which the content posted by the user and the input search query are registered as the log type is described. Various contents such as a dictionary and a web page are registered, and information indicating the type of each content is registered.

  Returning to FIG. 2, the description will be continued. In the distributed expression space database 32, the generated distributed expression space is registered. For example, FIG. 4 is a diagram illustrating an example of information registered in the distributed representation space database according to the embodiment. In the example illustrated in FIG. 4, information such as “distributed expression space ID”, “generation date / time”, “corresponding context”, “distributed expression”, and the like is registered in the distributed expression space database 32 in association with each other. In addition to the information shown in FIG. 4, for example, arbitrary information may be registered in the distributed representation space database 32 as long as the information is related to the distributed representation.

  Here, the “distributed expression space ID” is an identifier for identifying the distributed expression space. Further, the “generation date / time” is included in the distributed expression empty indicated by the associated “distributed expression space ID”, that is, the date and time when the distributed expression space indicated by the associated “distributed expression space ID” is generated. Information indicating the date and time when the distributed representation is generated. The “corresponding context” is information indicating a context that is a source of distributed representation. The “distributed expression” is a distributed expression of the associated “corresponding context”.

  For example, information such as a distributed expression space ID “space # 1”, a generation date “date and time # 4”, a corresponding context “A”, a distributed expression “distributed expression # 1”, and the like are registered in the distributed expression space database 32 in association with each other. Has been. Such information is a distributed expression generated on the date and time indicated by the generation date and time “date and time # 4” as the distributed expression included in the distributed expression space indicated by the distributed expression space ID “space # 1” and the corresponding context “ This indicates that the distributed expression of “A” is the distributed expression “distributed expression # 1”.

  In the example illustrated in FIG. 4, conceptual values such as “space ID # 1”, “date and time # 1”, “A”, and the like are described. Content such as a character string indicating a space or date and a text or an image as a context is registered.

  Returning to FIG. 2, the description will be continued. In the model database 33, a model for converting a context into a distributed representation is registered. For example, the model database 33 is a learning device that performs classification according to the similarity of each context, and the similarity of the user's impression with respect to the context (or the determination target indicated by the context) included in the log is one element. As a model, a learner that classifies each context is registered as a model. Such a model is realized by, for example, a learning device that classifies each context based on the appearance rate and similarity of other parts of speech such as adjectives that appear with the context. For example, the model may be realized by a multi-stage neural network such as a convolutional neural network (Convolutional Neural Networks: CNN). It may be realized.

  The control unit 40 is a controller. For example, various programs stored in a storage device inside the information providing apparatus 10 are stored in a RAM or the like by a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). Is implemented as a work area. The control unit 40 is a controller, and may be realized by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array). The control unit 40 is an arithmetic processing unit that reads an index.

  As illustrated in FIG. 2, the control unit 40 includes a collection unit 41, a generation unit 42, a reception unit 43, an acquisition unit 44, and a provision unit 45. The collection unit 41 collects various logs from an arbitrary information management server 100. For example, the collection unit 41 collects various logs such as content posted by the user, user action history, user search query history, information on an e-commerce target, information on an e-commerce web page, and the like. The collected logs are registered in the log database 31.

  The generation unit 42 generates a distributed representation of each context based on the relative relevance of a plurality of contexts. More specifically, the generation unit 42 generates a vector representing each context as a distributed representation using a model that vectorizes each context based on the relative relevance of a plurality of contexts.

  For example, the generation unit 42 extracts various logs posted during a predetermined period from various types of information registered as logs in the log database 31, and extracts a context from the extracted logs. Then, the generation unit 42 uses the model registered in the model database 33 to convert the context into a distributed representation based on the extracted context similarity. Thereafter, the generation unit 42 associates the date and time when the distributed expression is generated with the distributed expression space ID for the set of the generated distributed expression and the context that is the basis of the distributed expression, and registers them in the distributed expression space database 32. To do.

  The generation unit 42 generates a distributed representation from all the logs registered in the log database 31 at predetermined time intervals, and uses the generated distributed representation as a distributed representation space that is different for each generated timing. 32 may be registered. In addition, the generation unit 42 generates a plurality of distributed expressions from a log posted within one day from the date and time when the distributed expression is generated, and the generated plurality of distributed expressions are expressed as a single distributed expression space. It may be registered in the spatial database 32. That is, the generation unit 42 may generate a distributed expression space that reflects a user's impression in a predetermined period for each period.

  In addition, when a reference context is set in advance, the generation unit 42 may generate a distributed representation of each context so that the distributed representations of the reference context are separated from each other by a predetermined threshold or more. The generation unit 42 may generate a distributed expression of a plurality of contexts and predetermined adjectives.

  The accepting unit 43 accepts designation of a determination target context and a reference context from a consumer, and accepts designation of a date and time for determining a change in the determination target context. For example, the accepting unit 43 accepts designation of the date and time when the distribution information related to the determination target is distributed as the date and time when the change of the determination target context is determined. In the following description, the date and time when the change of the determination target context is determined is described as the designated date and time.

  The acquisition unit 44 acquires a distributed representation of the determination target context and the reference context before the specified date and time, and a distributed representation of the determination target context and the reference context after the specified date and time. For example, the acquisition unit 44 refers to the distributed representation space database 32 and selects a distributed representation space generated immediately before the specified date and a distributed expression space generated immediately after the specified date. In addition, the acquisition unit 44 acquires a distributed expression associated with the determination target context and a distributed expression associated with the reference context from each identified distributed expression space. That is, the acquisition unit 44 acquires a distributed representation of the determination target context generated before distribution information regarding the determination target context is distributed and a distributed representation of the determination target context generated after distribution information is distributed. To do. Then, the acquisition unit 44 outputs each identified distributed expression to the providing unit 45.

  The providing unit 45 indicates a change between a distributed representation of the determination target context generated before distribution information regarding the determination target context is distributed and a distributed representation of the determination target context generated after distribution information is distributed. Provide information. For example, the providing unit 45 includes a distributed representation of the determination target context generated before the advertisement information for advertising the determination target context is distributed, and a distributed representation of the determination target context generated after the advertisement information is distributed. Provide information indicating changes in Further, for example, the providing unit 45 generates a distributed representation of each context based on the relative relevance of an impression that an arbitrary user has with respect to a target indicated by each context.

  Here, the providing unit 45 generates information indicating a change in relevance between the distributed representation of the reference context and the distributed representation of the determination target context. More specifically, the providing unit 45 is generated after the difference between the distributed representation of the reference context generated before the distribution information is distributed and the distributed representation of the determination target context, and after the distribution information is distributed. The difference between the distributed representation of the reference context and the distributed representation of the determination target context is specified, and information indicating the change of the specified difference is generated. For example, the providing unit 45 generates information indicating a change in relevance between the distributed representation of the plurality of reference contexts and the distributed representation of the determination target context.

  Hereinafter, an example of processing performed by the providing unit 45 will be described. For example, the providing unit 45 is a distributed expression (hereinafter referred to as “distributed expression A”) indicating the context of the brand “A” as the distributed expression of the determination target context, and includes distribution information regarding the brand “A”. The distributed representation generated before distribution and the distributed representation generated after distribution are received from the acquisition unit 44. In addition, for example, the providing unit 45 may use a distributed expression (hereinafter, “distributed expression B”, “distributed expression C”) indicating the contexts of the brand “B”, the brand “C”, and the brand “E” as the distributed expression of the reference context. , And “distributed expression E”), and the distribution expression generated before distribution of the distribution information regarding the brand “A” and the distributed expression generated after distribution are received from the acquisition unit 44.

  In such a case, the providing unit 45 generates information indicating a change in the position where the determination target context is projected when each of the distributed representations of the reference context is projected to a predetermined position. For example, the providing unit 45 projects the distributed representation B, the distributed representation C, and the distributed representation E generated before distribution information on the brand “A” is distributed onto each vertex of an equilateral triangle having a predetermined size. A first normalization function is generated. Further, the providing unit 45 projects the distributed representation B, the distributed representation C, and the distributed representation E generated after distribution information on the brand “A” is distributed onto each vertex of a regular triangle of a predetermined size. A second normalization function is generated.

  Then, the providing unit 45 uses the first normalization function to place the distributed representation A generated before the distribution information is distributed on an equilateral triangle in which the distributed representation of the reference context is associated with each vertex. Project. Further, the providing unit 45 uses the second normalization function to project the distributed representation A generated after the distribution information is distributed onto an equilateral triangle in which the distributed representation of the reference context is associated with each vertex. To do. That is, when the distributed representation of the reference context is projected onto the vertexes of the polygon, the providing unit 45 generates information indicating the change in position in the region within the polygon where the determination target context is projected. For example, when the number of reference contexts is two, the providing unit 45 generates a normalization function that projects the reference context onto each vertex of the line segment, and uses the generated normalization function to determine the determination target context. May be generated on the line segment to generate information indicating a change in impression on the determination target.

  For example, FIG. 5 is a diagram illustrating an example of information generated by the information providing apparatus according to the embodiment. In the example illustrated in FIG. 5, as an example of information generated by the information providing apparatus 10, the change in impression of the brand “A” when the context between the brand “B” and the brand “E” is used as a reference context. An example of the content C20 indicating the change in the impression of the brand “A” when the content “C20” and the context of the brand “B”, the brand “C”, and the brand “E” are used as the reference context is described.

  First, an example of processing for generating content C20 will be described. For example, the information providing apparatus 10 projects the distributed representation E and the distributed representation B generated before the distribution information on the brand “A” is distributed onto the vertices of a predetermined length of the line segment. A function is generated, and using the first normalization function, the distribution information A generated before the distribution information is distributed is projected onto the line segment. Then, the information providing apparatus 10 arranges circular icons at the positions where the vertices and the distributed representation A are projected, and contents C21 in which the corresponding contexts “E”, “A”, and “B” are arranged are pre-delivery contents Generate as

  In addition, the information providing apparatus 10 projects a distributed expression E and a distributed expression B generated after distribution information on the brand “A” is distributed onto the vertices of a predetermined length of a line segment. And the distribution information A generated after the distribution information is distributed is projected onto the line segment using the second normalization function. Then, the information providing apparatus 10 arranges a circular icon at a position where each vertex and the distributed expression A are projected, and distributes the content C22 in which the corresponding contexts “E”, “A”, and “B” are arranged. Generate as

  In addition, the information providing apparatus 10 generates information indicating how close the distributed representation A is to the distributed representation B before and after the distribution information is distributed. For example, in the information providing apparatus 10, the distance (similarity) between the distributed representation A and the distributed representation B after distribution information distribution is about 30% of the distance between the distributed representation A and the distributed representation B before distribution information distribution. , Information indicating how close the distributed representation A is to the distributed representation B by the distribution of the distribution information, such as “achievement rate 70%”, is arranged in the content C22.

  Next, an example of processing for generating the content C30 will be described. For example, the information providing apparatus 10 generates a first normalization function that projects the distributed representation B, the distributed representation C, and the distributed representation E generated before the distribution information is distributed onto each vertex of the equilateral triangle, Using the first normalization function, the distribution information A generated before the distribution information is distributed is projected into an equilateral triangle. Then, the information providing apparatus 10 arranges a circular icon at a position where the distributed representation A is projected, and sets the contexts “E”, “B”, “C”, and “A” corresponding to each vertex and the circular icon. The arranged content C31 is generated as pre-distribution content.

  Further, the information providing apparatus 10 generates a second normalization function for projecting the distributed representation B, the distributed representation C, and the distributed representation E generated after the distribution information is distributed onto each vertex of the equilateral triangle. Using the normalization function of 2, the distribution information A generated after the distribution information is distributed is projected into an equilateral triangle. Then, the information providing apparatus 10 arranges a circular icon at a position where the distributed representation A is projected, and sets the contexts “E”, “B”, “C”, and “A” corresponding to each vertex and the circular icon. The arranged content C32 is generated as post-delivery content.

  Further, the information providing apparatus 10 has a case where the distance between the distributed representation A and the distributed representation B after distribution information distribution is about 30% of the distance between the distributed representation A and the distributed representation B before distribution information distribution. Information indicating how close the distributed representation A is to the distributed representation B by distributing the distribution information, such as “achievement rate 70%”, is arranged in the content C32. Note that the information providing apparatus 10 superimposes the pre-distribution content and the post-distribution content and arranges an arrow connecting the position before the movement of the distributed representation A and the position after the movement, thereby moving the distributed representation A. It may be shown visually.

  Here, the providing unit 45 may provide information indicating a history of changes in the position where the determination target context is projected. For example, when the distribution information is distributed a plurality of times, the providing unit 45 may generate information indicating a change in the distributed representation of the determination target context that occurs each time the distribution information is distributed.

  For example, FIG. 6 is a diagram illustrating an example of information indicating a history of changes in distributed representation provided by the information providing apparatus according to the embodiment. In the example illustrated in FIG. 6, an example of content C40 indicating a history of changes in impression of brand “A” when the context of brand “B”, brand “C”, and brand “E” is used as a reference context. Described. Further, in the example illustrated in FIG. 6, an example of the content C40 generated when the change in the impression of the user with respect to the brand “A” is determined in four times is described.

For example, the information providing apparatus 10 uses the first normalization function for projecting the distributed representation B, the distributed representation C, and the distributed representation E generated at the first timing on each vertex of the equilateral triangle, and uses the first timing. Project the distributed representation A1 generated in ( ₁ ). Similarly, the information providing apparatus 10 projects the distributed representation B, the distributed representation C, and the distributed representation E generated at the second timing to the fourth timing onto each vertex of the equilateral triangle. Using the normalization function of 4, the distributed expressions A _{2 to} A ₄ generated at the second timing to the fourth timing are projected. Then, as shown in FIG. 6, the information providing apparatus 10 arranges an arrow connecting the distributed expressions A _{1 to} A ₄ , that is, an arrow indicating the transition of the user's impression with respect to the brand “A”. As a result, the information providing apparatus 10 can indicate how the impression of the user with respect to the brand “A” has changed as compared with the reference context by the distribution information.

  In the example illustrated in FIG. 6, the first timing to the fourth timing may or may not correspond to the distribution timing of the distribution information corresponding to the brand “A”. For example, each time the distribution information corresponding to the brand “A” is distributed, the information providing apparatus 10 executes the above-described process, so that the distribution information distributed at each timing induces the user's impression. A content C40 indicating the above may be generated. In addition, for example, the information providing apparatus 10 distributes the distribution information corresponding to the brand “A” once, and then performs the above-described processing at a predetermined time interval for a predetermined period, so that the distribution information is obtained. The content C40 indicating how the user's impression is induced may be generated.

  The providing unit 45 may use a context belonging to a category common to the determination target context as a reference context. For example, when only the designation of the determination target context is accepted, the providing unit 45 may automatically select a category context common to the specified determination target context as the reference context.

  The providing unit 45 may exclude information indicating a distributed representation of contexts other than the reference context and the determination target context from the plurality of contexts from the provision target. For example, when providing the drawing in which the distributed representation space is projected on a predetermined plane, the providing unit 45 may generate a drawing in which only the reference context and the determination target context are projected.

  In addition, when a predetermined adjective is included as an anchor in the distributed expression space, the providing unit 45 generates information indicating a change in the relationship between the distributed expression of the predetermined adjective and the distributed expression of the determination target context. May be. For example, when the consumer desires information indicating whether or not the distribution information has approximated the impression of the brand “A” to “high class”, information indicating the position of the distributed expression A relative to the anchor “high class” is generated. May be.

  Further, the providing unit 45 may evaluate the distribution information based on a change in the distributed representation of the determination target context, and generate an evaluation result of the distribution information as information indicating the change in the distributed representation. For example, when the consumer desires information indicating whether or not the distribution information approximates the impression of the brand “A” to the brand “B”, the providing unit 45 may make the relative similarity between the distributed expression A and the distributed expression B. The change in the degree is specified, and the distribution information is evaluated according to the change in the specified similarity. To give a more specific example, if the increase rate of the relative similarity between the distributed representation A and the distributed representation B exceeds a predetermined threshold after the distribution information is distributed, the providing unit 45 An evaluation result indicating that the impression of the brand “A” is brought close to the brand “B” is generated. On the other hand, if the increase rate of the relative similarity between the distributed representation A and the distributed representation B does not exceed a predetermined threshold or does not change after the distribution information is distributed, the providing unit 45 If the degree of relative similarity with the distributed representation B decreases, an evaluation result indicating that the distribution information did not bring the impression of the brand “A” close to the brand “B” is generated.

  Then, the providing unit 45 distributes the generated various types of information to the consumer server 300. As a result, the information providing apparatus 10 can provide information indicating the influence of the distribution information on the impression of the user with respect to the determination target, so that the evaluation of the influence of the distribution information can be realized.

[4. Example of flow of processing executed by information providing apparatus]
Next, the flow of providing processing executed by the information providing apparatus 10 will be described with reference to FIG. FIG. 7 is a flowchart illustrating an example of a flow of providing processing executed by the information providing apparatus according to the embodiment. In addition, the information provision apparatus 10 can perform the process shown in FIG. 7 by arbitrary units at arbitrary timings.

  For example, the information providing apparatus 10 generates a distributed expression from log data before distribution information is distributed based on the relative relevance of each context (step S101). Subsequently, the information providing apparatus 10 generates a distributed expression based on the relative relevance of each context from the log data after the distribution information is distributed (step S102). Then, the information providing apparatus 10 generates a normalization function for projecting the distributed representation of the reference context to a predetermined position for each of the distribution information before and after (step S103).

  Further, the information providing apparatus 10 generates content indicating a change in the positional relationship between the determination target context and the reference context (step S104). Then, the information providing apparatus 10 provides the generated content as the evaluation result of the distribution information (Step S105), and ends the process.

[5. (Modification)
In the above, an example of the providing process and the calculating process by the information providing apparatus 10 has been described. However, the embodiment is not limited to this. Hereinafter, variations of the providing process and the calculation process executed by the information providing apparatus 10 will be described.

[5-1. About the evaluation target)
In the example described above, the information providing apparatus 10 evaluates distribution information related to an arbitrary determination target indicated by the context, such as advertisement information. Here, the information providing apparatus 10 may provide information for evaluating arbitrary distribution information. For example, the information providing apparatus 10 may provide information for evaluating arbitrary distribution information such as news, a web page, and a post format advertisement (that is, a native advertisement). In addition, the information providing apparatus 10 may provide, for example, a change in the user's impression of a company that sells a product before and after the sale of the product.

  In other words, the information providing apparatus 10 generates not only the evaluation of various information transmitted to the user via the network but also before the occurrence of a predetermined event related to the determination target context, such as sales of products or provision of events. By providing information indicating a change between the distributed representation of the determined determination target context and the distributed representation of the determination target context generated after the predetermined event related to the determination target context occurs, the event is added to the impression of the user. It is possible to appropriately evaluate the impacts.

[5-2. Device configuration〕
The information providing apparatus 10 may be realized by a front end server and a back end server. In such a case, the reception unit 43 and the provision unit 45 shown in FIG. 2 are arranged in the front-end server, and the collection unit 41, the generation unit 42, and the acquisition unit 44 are arranged in the back-end server. The databases 31 to 33 registered in the storage unit 30 may be held in an external storage server.

[5-3. Others]
In addition, among the processes described in the above embodiment, all or part of the processes described as being automatically performed can be performed manually, and conversely, the processes described as being performed manually. All or a part of the above can be automatically performed by a known method. In addition, the processing procedures, specific names, and information including various data and parameters shown in the document and drawings can be arbitrarily changed unless otherwise specified. For example, the various types of information illustrated in each drawing is not limited to the illustrated information.

  Further, each component of each illustrated apparatus is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured.

  In addition, the above-described embodiments can be appropriately combined within a range in which processing contents do not contradict each other.

[5-4. program〕
Further, the information providing apparatus 10 according to the above-described embodiment is realized by a computer 1000 having a configuration as shown in FIG. FIG. 8 is a diagram illustrating an example of a hardware configuration. The computer 1000 is connected to an output device 1010 and an input device 1020, and an arithmetic device 1030, a primary storage device 1040, a secondary storage device 1050, an output IF (Interface) 1060, an input IF 1070, and a network IF 1080 are connected via a bus 1090. Have

  The arithmetic device 1030 operates based on a program stored in the primary storage device 1040 and the secondary storage device 1050, a program read from the input device 1020, and the like, and executes various processes. The primary storage device 1040 is a memory device such as a RAM that temporarily stores data used by the arithmetic device 1030 for various arithmetic operations. The secondary storage device 1050 is a storage device in which data used by the arithmetic device 1030 for various calculations and various databases are registered, such as ROM (Read Only Memory), HDD (Hard Disk Drive), flash memory, and the like. It is realized by.

  The output IF 1060 is an interface for transmitting information to be output to an output device 1010 that outputs various types of information such as a monitor and a printer. For example, USB (Universal Serial Bus), DVI (Digital Visual Interface), This is realized by a standard connector such as HDMI (registered trademark) (High Definition Multimedia Interface). The input IF 1070 is an interface for receiving information from various input devices 1020 such as a mouse, a keyboard, and a scanner, and is realized by, for example, a USB.

  The input device 1020 includes, for example, an optical recording medium such as a CD (Compact Disc), a DVD (Digital Versatile Disc), and a PD (Phase change rewritable disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), and a tape. It may be a device that reads information from a medium, a magnetic recording medium, a semiconductor memory, or the like. The input device 1020 may be an external storage medium such as a USB memory.

  The network IF 1080 receives data from other devices via the network N and sends the data to the arithmetic device 1030, and transmits data generated by the arithmetic device 1030 to other devices via the network N.

  The arithmetic device 1030 controls the output device 1010 and the input device 1020 via the output IF 1060 and the input IF 1070. For example, the arithmetic device 1030 loads a program from the input device 1020 or the secondary storage device 1050 onto the primary storage device 1040, and executes the loaded program.

  For example, when the computer 1000 functions as the information providing device 10, the arithmetic device 1030 of the computer 1000 implements the function of the control unit 40 by executing a program loaded on the primary storage device 1040.

[6. effect〕
As described above, the information providing apparatus 10 generates a distributed representation of each context based on the relative relevance of a plurality of contexts. Then, the information providing apparatus 10 includes a distributed representation of the determination target context generated before the distribution information regarding the determination target context is distributed, and a distributed representation of the determination target context generated after the distribution information is distributed. Provide information indicating changes. For this reason, the information providing apparatus 10 can provide information for evaluating the influence of the distribution information on the user's impression on the determination target.

  Further, the information providing apparatus 10 provides information indicating a change in relevance between the distributed representation of the reference context and the distributed representation of the determination target context. For example, the information providing apparatus 10 determines the difference between the distributed representation of the reference context generated before the distribution information is distributed and the distributed representation of the determination target context, and the reference context generated after the distribution information is distributed. Information indicating a change between the distributed representation and the difference between the distributed representation of the determination target context is provided. In addition, the information providing apparatus 10 provides information indicating a change in relevance between the distributed representation of the plurality of reference contexts and the distributed representation of the determination target context. For this reason, the information providing apparatus 10 can provide a change in the relative relevance of the determination target context with respect to the reference context even when the impression of the user with respect to various contexts changes sequentially. Information for evaluating the influence of the distribution information on the user's impression of the target can be provided.

  Further, the information providing apparatus 10 provides information indicating a change in the position where the determination target context is projected when each of the distributed representations of the plurality of reference contexts is projected to a predetermined position. For example, the information providing apparatus 10 projects the first function for projecting the distributed representation of the reference context before the distribution information is distributed to a predetermined position and the distributed expression of the reference context after the distribution information is distributed. A second function to be projected to the position of the first position, and a position where the distributed representation of the determination target context before the distribution information is distributed is projected by the first function, and a determination target after the distribution information is distributed Information indicating a distributed representation of the context and the position projected by the second function is provided. For this reason, the information provision apparatus 10 can produce | generate and provide the information which shows appropriately the influence which distribution information exerted on the user's impression with respect to a consumer.

  Further, the information providing apparatus 10 provides information indicating a change in position in an area on the straight line or in the polygon on which the determination target context is projected when the distributed representation of the reference context is projected onto the vertex of the straight line or the polygon. To do. For this reason, the information provision apparatus 10 can produce | generate and provide the information which shows visually the influence which distribution information had on a user's impression with respect to a consumer.

  Further, the information providing apparatus 10 generates a distributed representation of each context so that the distributed representations of the reference context are separated from each other by a predetermined threshold or more. For this reason, since the information provision apparatus 10 can emphasize the change of the user's impression with respect to a determination object context, it can provide the information which is easy to evaluate the influence which distribution information exerted on the user's impression.

  Further, the information providing apparatus 10 sets a context belonging to a category common to the determination target context as a reference context. For this reason, the information providing apparatus 10 can provide information for appropriately evaluating the influence of the distribution information on the user's impression.

  In addition, the information providing apparatus 10 excludes information indicating a distributed representation of contexts other than the reference context and the determination target context from the plurality of contexts from the provision target. For this reason, the information providing apparatus 10 can provide information that makes it easy to evaluate the influence of the distribution information on the user's impression.

  Further, the information providing apparatus 10 provides information indicating a history of changes in the distributed representation of the determination target context. For this reason, since the information providing apparatus 10 can provide the user with a history of changes in the user's impression on the determination target, the information providing apparatus 10 provides information for appropriately evaluating the influence of the distribution information on the user's impression. it can.

  Further, the information providing apparatus 10 evaluates the distribution information based on the change in the distributed representation of the determination target context, and provides the evaluation result of the distribution information as information indicating the change in the distributed representation. For this reason, the information provision apparatus 10 can provide the evaluation result which shows whether the delivery information induced the user's impression appropriately, for example.

  Moreover, the information provision apparatus 10 produces | generates the vector which shows each context as a distributed expression using the learning device which vectorizes each context based on the relative relationship which a some context has. In addition, the information providing apparatus 10 includes information including at least one of content posted by an arbitrary user, an action history of an arbitrary user, a search history of an arbitrary user, or information on an electronic commerce target. To generate a distributed representation of each context. Moreover, the information provision apparatus 10 produces | generates the distributed expression of each context based on the relative relevance of the impression which an arbitrary user has with respect to the object which each context shows. For this reason, since the information providing apparatus 10 provides information indicating the change in the distributed expression reflecting the user's impression, the information providing apparatus 10 can provide information for appropriately evaluating the influence of the distribution information on the user's impression.

  Further, the information providing apparatus 10 generates a distributed representation of a plurality of contexts and a predetermined adjective, and provides information indicating a change in the relationship between the distributed representation of the predetermined adjective and the distributed representation of the determination target context. . For this reason, the information providing apparatus 10 can provide information indicating how the user's impression on the determination target has changed using a predetermined adjective as an anchor.

  Further, the information providing apparatus 10 may provide information indicating a change between a distributed expression generated before a predetermined event related to the determination target context occurs and a distributed expression generated after the event occurs. As a result of such processing, the information providing apparatus 10 can provide a change in the distributed expression before and after the occurrence of an event related to the determination target such as the start of provision of goods and services, a brand, a company, an event hosted by a region, etc. It is possible to provide information for evaluating whether an event such as an event has appropriately induced the user's impression.

  Further, the information providing apparatus 10 includes a distributed representation of the determination target context generated before the advertisement information for advertising the determination target context is distributed, and the determination target context generated after the advertisement information is distributed. Provides information indicating changes to the distributed representation of For this reason, the information providing apparatus 10 can provide information for evaluating whether the advertisement information appropriately induces the user's impression of the determination target.

  As described above, some of the embodiments of the present application have been described in detail with reference to the drawings. However, these are merely examples, and various modifications, including the aspects described in the disclosure section of the invention, based on the knowledge of those skilled in the art, It is possible to implement the present invention in other forms with improvements.

  Moreover, the above-mentioned “section (module, unit)” can be read as “means”, “circuit”, and the like. For example, the providing unit can be read as providing means or a providing circuit.

DESCRIPTION OF SYMBOLS 10 Information provision apparatus 20 Communication part 30 Storage part 31 Log database 32 Distributed expression space database 33 Model database 40 Control part 41 Collection part 42 Generation part 43 Reception part 44 Acquisition part 45 Providing part 100 Information management server 200 Distribution server 300 Consumer server 400 User terminal

Claims (17)

A distributed representation of the context that reflects the user's impression indicated by the text based on content including text indicating the user's impression of the context, and a relative expression that relatively indicates the user's impression. A generation unit that generates each context of the same type;
Generated after the occurrence of the given event and the relationship between the distributed representation of the given context generated before the given event for the given context occurs and the distributed representation of the plurality of reference contexts. And a providing unit that provides information indicating a change in the relationship between the distributed representation of the predetermined context and the distributed representation of the plurality of reference contexts. The providing apparatus according to claim 1, wherein the providing unit provides information indicating a change in relevance between a distributed representation of a reference context and the distributed representation of the predetermined context. The providing unit includes a difference between the distributed representation of the reference context generated before the predetermined event occurs and the distributed representation of the predetermined context, and the generated generated after the predetermined event occurs. The providing apparatus according to claim 2, wherein information indicating a change between a distributed representation of a reference context and a difference between the distributed representation of the predetermined context is provided. The said provision part provides the information which shows the change of the relationship between the distributed expression of the some reference | standard context and the distributed expression of the said predetermined context, The any one of Claims 1-3 characterized by the above-mentioned. The providing device according to one. The said provision part provides the information which shows the change of the position where the said predetermined context is projected, when the distributed expression of the some reference | standard context is each projected on the predetermined position. The providing device according to 1. The providing device according to any one of claims 1 to 5, wherein the providing unit sets a context belonging to a category common to the predetermined context as the reference context. The said provision part excludes the information which shows the distributed expression of contexts other than the reference | standard context and the said predetermined context from the provision object among the said contexts, The any one of Claims 1-6 characterized by the above-mentioned. The providing device according to 1. The providing device according to any one of claims 1 to 7, wherein the providing unit provides information indicating a history of changes in the distributed representation of the predetermined context. The generation unit generates a vector indicating each context as the distributed representation using a learning device that vectorizes each context based on a relative relevance of a plurality of contexts. The providing apparatus according to any one of 8. Each of the generation units includes, from information including at least one of content posted by an arbitrary user, behavior history of an arbitrary user, search history of an arbitrary user, or information on an object of electronic commerce, The providing apparatus according to claim 9, wherein a distributed representation of the context is generated. The generation unit generates a distributed representation of a plurality of contexts and predetermined adjectives,
The said provision part provides the information which shows the change of the relationship between the distributed expression of a predetermined adjective, and the distributed expression of the said predetermined context. The one of Claims 1-10 characterized by the above-mentioned. Providing device. The providing unit includes, as the predetermined event, a distributed representation of the predetermined context generated before distribution information relating to the predetermined context is distributed, and the predetermined event generated after the distribution information is distributed. The information which shows the change with the distributed expression of context of this is provided. The provision apparatus as described in any one of Claims 1-11 characterized by the above-mentioned. The providing unit evaluates the distribution information based on a change in the distributed representation of the predetermined context, and provides an evaluation result of the distribution information as information indicating the change in the distributed representation. Item 13. The providing device according to Item 12. The providing unit generates, as the predetermined event, a distributed representation of the predetermined context generated before the advertising information for advertising the predetermined context is distributed, and after the advertising information is distributed The information which shows the change with the distributed expression of the said predetermined context is provided. The provision apparatus as described in any one of Claims 1-13 characterized by the above-mentioned. The generation unit generates a distributed representation of each context based on a relative relevance of an impression of an arbitrary user with respect to a target indicated by each context. The providing device according to any one of the above. A providing method executed by a providing device,
A distributed representation of the context that reflects the user's impression indicated by the text based on content including text indicating the user's impression of the context, and a relative expression that relatively indicates the user's impression. A generation process for generating each kind of context;
Generated after the occurrence of the given event and the relationship between the distributed representation of the given context generated before the given event for the given context occurs and the distributed representation of the plurality of reference contexts. A providing step of providing information indicating a change between the distributed representation of the predetermined context and the relationship between the distributed representation of the plurality of reference contexts. A distributed representation of the context that reflects the user's impression indicated by the text based on content including text indicating the user's impression of the context, and a relative expression that relatively indicates the user's impression. Generation procedure to generate for each kind of context,
Generated after the occurrence of the given event and the relationship between the distributed representation of the given context generated before the given event for the given context occurs and the distributed representation of the plurality of reference contexts. A providing program for causing a computer to execute a providing procedure for providing information indicating a change between the distributed representation of the predetermined context and the relationship between the distributed representation of the plurality of reference contexts.

Images