JP5887031B1 - Product identification device, product identification method, and product identification program - Google Patents

Abstract

The product specifying device is a product page acquisition unit that acquires a product page to be processed, and the product page includes text representing the product related to the product page and can be divided into tokens including one or more characters. A product page acquisition means including a page title, and an acquisition means for acquiring at least one preset product name, wherein the product name includes a product name acquisition means including one or more tokens, and a product page title of the product page If the estimation means for estimating the feature relating to the number of tokens included in the feature and the feature estimated by the estimation means indicate that the number of tokens is less than a predetermined level, the first processing suitable for such a case is executed, When the estimated feature indicates that the number of tokens is equal to or greater than a predetermined level, execution means for executing a second process suitable for such a case, Comprising processing or in accordance with the calculated scores by the second processing, a specifying means for specifying a product name identity with product pages and product.

Description

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

  Conventionally, a technique for determining the identity of two text data contents is known. For example, Patent Document 1 describes a technique for determining the identity of two text data by paying attention to a conversion operation for matching one text data with the other text data.

JP 2010-134501 A

In an electronic commerce site where different stores can sell the same product, the same product may need to be associated with each other. For example, a product page of a product and a product page of the same product are displayed in association with each other, or a thumbnail of a product page indicating the same product as the searched or specified product is included in one page and displayed. . Thus, it is easy to associate the same products with each other if there is a product code that is common among a plurality of stores.
However, if there is a product that is not associated with a common product code, it is necessary to collate the official product name list with the product name handled on the product page. On the other hand, the product page on the electronic commerce site may include various words for so-called SEO (Search Engine Optimization) measures to be displayed at the top in the search results by the search engine. In order to determine the identity of the product name shown on the product page of the electronic commerce site and the official product name, a word for so-called SEO (so-called spam word) is unnecessary in the identity determination process by using a dictionary. It is possible to exclude words. However, if a product page contains many words that are not spam words, it is not possible to sufficiently exclude words that are not necessary for identity determination, and there is a product page that should be judged to be the same as the official product name. In some cases, it was not determined to be the same.

  Therefore, an object of the present invention is to more accurately determine the identity between the product name indicated on the product page and the official product name.

  In order to solve the above-described problem, a product specifying device according to an aspect of the present invention is a product page acquisition unit that acquires a product page to be processed, and the product page is based on text representing a product related to the product page. A product page acquisition means including a product page title that can be divided into one or more tokens, and a product name acquisition means for acquiring at least one preset product name, wherein the product name is one or more Product name acquisition means including one or more tokens made up of characters, estimation means for estimating the characteristics relating to the number of tokens included in the product page title of the product page, and the characteristics estimated by the estimation means, The first process is executed when it indicates that the number is low, and the second process is executed when the feature indicates that the number of tokens is equal to or greater than a predetermined level. The first processing is a first process in which the ratio of the other of the token included in the product name and the token included in the product page title is compared with the product indicated by the product name and the product related to the product page. The second process calculates a ratio in which one of the token included in the product name and the token included in the product page title is included in the other as a score indicating the identity. , An executing means, and a specifying means for specifying a product name having the same identity as the product page and the product according to the score calculated by the first process or the second process.

  A product specifying method according to an aspect of the present invention is a product specifying method in a product specifying device, and is a product page acquiring step of acquiring a product page to be processed, wherein the product page is a product page related to the product page. A product page acquisition step including a product page title that includes a product page title that is composed of a text representing and can be divided into one or more tokens, and a product name acquisition step of acquiring at least one preset product name, wherein the product name is: A product name acquisition step including one or more tokens including one or more characters, an estimation step for estimating a feature related to the number of tokens included in the product page title of the product page, and a feature estimated in the estimation step is the number of tokens If it indicates that the number is less than the predetermined level, the first process is executed, and the feature is that the number of tokens is equal to or higher than the predetermined level. The execution step of executing the second process in the case of indicating that there is a ratio, wherein the first process is a ratio of the other match to one of the token included in the product name and the token included in the product page title. Is calculated as a score indicating the identity between the product indicated by the product name and the product related to the product page, and the second process is such that one of the token included in the product name and the token included in the product page title is the other The product name having the identity of the product page and the product is identified according to the execution step of calculating the ratio included in the score as a score indicating identity and the score calculated by the first process or the second process. Specific steps to be performed.

  A product specifying program according to an aspect of the present invention is a product page acquisition function for acquiring a product page to be processed in a computer, and the product page includes text representing the product related to the product page and at least one product page. A product page acquisition function including a product page title that can be divided into tokens, and a product name acquisition function that acquires at least one preset product name, wherein the product name is a token composed of one or more characters. When the product name acquisition function including one or more, the estimation function for estimating the feature regarding the number of tokens included in the product page title of the product page, and the feature estimated by the estimation function indicate that the number of tokens is less than a predetermined level If the feature indicates that the number of tokens is greater than or equal to a predetermined level, execute the second process. In the first process, the ratio of the other of the token included in the product name and the token included in the product page title is matched between the product indicated by the product name and the product related to the product page. Calculate as a score indicating identity, and the second process calculates, as a score indicating identity, a ratio in which one of the token included in the product name and the token included in the product page title is included in the other. The execution function and a specific function for specifying a product name having the same product as the product page are realized according to the score calculated by the first process or the second process.

  According to said form, the characteristic regarding the token number contained in the merchandise page title of a merchandise page is estimated, and the goods and goods which a merchandise name shows by 1st process or 2nd process according to the estimated characteristic. A process of calculating a score indicating the identity with the product related to the page is executed. The first process is a process in which, when the number of tokens included in the product page title is small, a score that appropriately reflects the identity between the product indicated by the product name and the product related to the product page is calculated. On the other hand, the second process is a process in which, when the number of tokens included in the product page title is large, a score that appropriately reflects the identity between the product indicated by the product name and the product related to the product page is calculated. . The first process is performed when the characteristic regarding the estimated number of tokens indicates that the number of tokens is less than a predetermined level, and the second process is performed when the number of tokens indicates that the number of tokens is equal to or greater than the predetermined level. Is executed, a score that appropriately reflects the identity is calculated according to the product page to be processed. Since the score appropriately reflecting the identity is used, it is possible to specify the product name having the same identity as the product page with high accuracy.

  In the product identification device according to another aspect, the product name acquisition unit acquires a product name preset by a person other than the store handling the product related to the processing target product page, and the estimation unit is input by the store According to the feature of one or more product pages, the feature related to the token number of the product page title of the product page to be processed is estimated.

  In the said form, since the goods name preset by persons other than a store is acquired, the goods name suitable for judgment of the identity with a goods page is acquired. In addition, the product page used for estimating the characteristics related to the number of tokens of the product page title of the product page to be processed is a product page input by a store handling the product related to the product page, and the product page to be processed is This is a product page that is included and is not limited to the product page to be processed. Since such a product page is suitable for estimating a feature related to the number of tokens of a product page to be processed, a feature related to the number of tokens is estimated with high accuracy.

  In the product identification device according to yet another aspect, the estimation unit is configured to determine the product page of the product page to be processed based on the amount of data constituting the product page input by the store handling the product related to the product page to be processed. It is good also as estimating the characteristic regarding the token number of a title.

  The characteristic regarding the token number of the product page title of the product page to be processed is reflected in the data amount constituting the product page input by the store handling the product related to the product page to be processed. According to the above aspect, since the feature regarding the token number of the product page title of the product page to be processed is estimated based on the data amount, the reliability of the estimated feature is improved.

  In the product identification device according to another aspect, in the above aspect, the estimation unit may determine that the number of tokens of the product page title of the product page to be processed is equal to or greater than a predetermined level when the data amount is equal to or greater than the predetermined amount. If the estimated amount of data is less than a predetermined amount, it may be estimated that the number of tokens of the product page title of the product page to be processed is less than a predetermined level.

  According to the said form, the characteristic regarding the token number of the goods page title of the goods page to be processed is estimated appropriately.

  In the product identification device according to another aspect, the estimation unit is configured to determine the product page of the product page to be processed based on the number of words of the text that configures the product page input by the store handling the product related to the product page to be processed. It is good also as estimating the characteristic regarding the token number of a title.

  The feature related to the token number of the product page title of the product page to be processed is reflected in the number of text words constituting the product page input by the store handling the product related to the product page to be processed. According to the above aspect, since the feature related to the token number of the product page title of the product page to be processed is estimated based on the number of words, the reliability of the estimated feature is improved.

  In the product identification device according to another aspect, the estimation unit is configured to process the product page to be processed based on the token number of the product page title included in the product page input by the store that handles the product related to the product page to be processed. It is good also as estimating the characteristic regarding the token number of the product page title.

  The feature regarding the number of tokens of the product page title of the product page to be processed is reflected in the number of tokens included in the product page title of the product page input by the store handling the product related to the product page to be processed. According to the above aspect, since the feature related to the token number of the product page title of the product page to be processed is estimated based on the number of tokens, the reliability of the estimated feature is improved.

  In the product identification device according to another aspect, the estimation unit has a predetermined ratio of the number of tokens of the product page title included in the product page to be processed with respect to the number of tokens included in the product name acquired by the product name acquisition unit. If the percentage is greater than or equal to the percentage, the number of tokens in the product page title of the product page to be processed is estimated to be greater than the predetermined level, and if the percentage is less than the predetermined percentage, the token of the product page title of the product page to be processed It may be estimated that the number is less than a predetermined level.

  According to the above embodiment, the number of tokens of the product page title included in the product page to be processed with respect to the number of tokens included in the product name, which directly reflects the characteristics related to the number of tokens of the product page title of the product page to be processed. Since the feature related to the number of tokens is estimated based on the ratio of, the reliability of the estimated feature can be improved.

  In the product identification device according to another aspect, the product name acquisition unit may acquire a product name including at least one token of a product page title included in the product page to be processed.

  According to the above embodiment, the number of product names used for processing can be limited by setting the product name including the token of the product page title included in the product page to be processed, thereby reducing the processing load. Is done.

  According to one aspect of the present invention, it is possible to more accurately determine the identity between the product name indicated on the product page and the official product name.

It is a block diagram which shows the function structure of a goods specific device. It is a figure which shows the hardware constitutions of a goods specific device. It is a figure which shows the example of the data memorize | stored in the merchandise name memory | storage part. It is a figure which shows the example of a goods page. It is a figure which shows the example of the some token obtained by dividing | segmenting the product page title of a product page. It is a figure which shows the example of the brand name acquired in order to use for a process. It is a figure which shows the score which shows the identity with each goods name calculated about the goods page title shown in FIG. It is a figure which shows the example of the some token obtained by dividing | segmenting the product page title of a product page. It is a figure which shows the score which shows the identity with each goods name calculated about the goods page title shown in FIG. It is a flowchart which shows the processing content of the goods specific method implemented in a goods specific device. It is a figure which shows the structure of a goods specific program.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same or equivalent elements are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a block diagram showing a functional configuration of a product identification device 1 according to the present embodiment. The product specifying device 1 is a device for specifying a product name having the same product item as the product page. The product page is, for example, a web page that presents a product on an electronic commerce site, and includes a product page title.

  Specifically, the product identification device 1 determines the identity of a product page title included in the product page to be processed and a preset official product name. Thereby, the official product name determined to have the same identity as the product page title is specified as the product name having the same identity as the product page and the product. That is, the product indicated by the product page is the same as the product indicated by the specified product name.

  The product page title included in the product page includes text representing the product related to the product page. Specifically, since the product page title includes text representing the product set by the store that handles the product, the text may include the product name of the product. Therefore, the product page title may include a word indicating an official product name and other words. In addition, the product page title may include various words for SEO countermeasures in addition to the word representing the official product name, for example. In this way, the product page title including various words can be divided into tokens by tokenizing (sharing).

  This token is obtained, for example, by dividing the product page title according to a predetermined rule such as a blank as a boundary. The token is composed of one or more characters and may be a single word or a collection of characters having no meaning.

  As described above, the product name is an official name preset for the product. The product name includes one or more tokens. Accordingly, the product page title of the product page often includes a token included in the official product name of the product indicated by the product page and a token not included in the official product name.

  Since the product identification device 1 of the present embodiment can identify the official product name having the same identity as the product name indicated by the product page, for example, in the electronic commerce site, the product page related to the same product is accurately displayed. Can be displayed in association. For example, in association with one product page designated by the user or one product page searched, thumbnails of a plurality of product pages indicating the same product can be included in one page and displayed.

  As shown in FIG. 1, the product identification device 1 functionally includes a product page acquisition unit 11 (product page acquisition unit), a product name acquisition unit 12 (product name acquisition unit), an estimation unit 13 (estimation unit), An execution unit 14 (execution means) and a specification unit 15 (specification means) are provided.

  The product identification device 1 can access storage means such as a product page storage unit 21 and a product name storage unit 22. The product page storage unit 21 and the product name storage unit 22 may be provided in the product specifying device 1 or may be configured as an external storage unit provided so as to be accessible from the product specifying device 1.

  FIG. 2 is a hardware configuration diagram of the product identification device 1. As shown in FIG. 2, the product identification device 1 physically includes a CPU 101, a main storage device 102 composed of a memory such as a RAM and a ROM, an auxiliary storage device 103 composed of a hard disk, a communication control device 104, and the like. It is comprised as a computer system containing. The product identification device 1 may further include an input device 105 such as a keyboard, a touch panel, and a mouse, which are input devices, and an output device 106 such as a display.

  The functions shown in FIG. 1 operate the communication control device 104 and the like under the control of the CPU 101 by reading predetermined computer software on the hardware such as the CPU 101 and the main storage device 102 shown in FIG. In addition, it is realized by reading and writing data in the main storage device 102 and the auxiliary storage device 103. Data and databases necessary for processing are stored in the main storage device 102 and the auxiliary storage device 103.

  Referring back to FIG. 2, the product page storage unit 21 and the product name storage unit 22 will be described prior to the description of each functional unit of the product identification device 1. The product page storage unit 21 is storage means for storing product page data. This product page is a page displayed on an electronic commerce site, and represents a product handled by one or more stores that sell products on the electronic commerce site.

  FIG. 3 is a diagram illustrating an example of a product page displayed based on data stored in the product page storage unit 21. The product page M includes a product page title T. The product page M may further include a product image G and a product description H. As described above, the product page title T includes text representing the product related to the product page, and includes a word indicating an official product name and a word indicating other than the official product name. The product page title T is attached with a predetermined tag indicating, for example, a title or a heading in the product page data. Therefore, by searching for the tag, the product page title is obtained from the product page data. It is possible to extract T. Further, in the product page data, text having a font size larger than other texts or text expressed in bold may be extracted as the product page title T.

  The product name storage unit 22 is a storage unit that stores an official product name of a product handled in the electronic commerce site to which the product specifying device 1 belongs. The official product name may be set and registered in advance by an administrator of the electronic commerce site and a store belonging to the electronic commerce site.

  FIG. 4 is a diagram illustrating an example of product names stored in the product name storage unit 22. As illustrated in FIG. 4, various product names such as “brandA abc08-bk eco” are stored in the product name storage unit 22.

Next, each functional unit of the product identification device 1 will be described. The product page acquisition unit 11 is a part that acquires a product page to be processed. An example of the product page acquired by the product page acquisition unit 11 is as shown in FIG. FIG. 5 is a diagram illustrating an example of the product page title extracted from the product page M to be processed. That is, product page title T ₁ shown in FIG. 5 is a text obtained by extracting the product page title T included in the product page M shown in FIG. Product Page Title _{T 1,} as shown in FIG. 5, can be divided into a plurality of tokens _t 11 _{~t 15.}

  In addition to the product page to be processed, the product page acquisition unit 11 may further acquire product pages other than the product page to be processed, which are input by a store that handles products related to the product page to be processed. . Specifically, since the product page to be processed is associated with the identifier of the store that handles the product related to the product page, the product page acquisition unit 11 stores the product page having the identifier of the store in the product page. It can be extracted from the unit 21.

  The product name acquisition unit 12 is a part that acquires one or more product names. Specifically, the product name acquisition unit 12 acquires the official product name stored in the product name storage unit 22. The product name acquired by the product name acquisition unit 12 is a product name set by a person other than the store, unlike the product name set as the product page title by the store handling the product related to the product page. The formal name preset for the product by the administrator of the electronic commerce site. The product name acquired by the product name acquisition unit 12 includes one or more tokens.

The product name acquisition unit 12 may acquire a product name including at least one product page title token included in the product page to be processed. For example, a product page M of the obtained processed by the commodity page acquisition unit 11, if the item page titles T shown in FIG. 5 is extracted, trade name acquisition unit 12, either the token t ₁₁ ~t ₁₅ The product name including the above is extracted from the product name storage unit 22. 6, taken from trade name storage unit 22 by the trade name acquisition unit 12 is a diagram showing the trade name n1~n4 containing either the token _t 11 _{~t 15.}

For example, the product name n1 includes tokens t ₁₁ “brandA”, t ₁₂ “abc08-bk”, and t ₁₃ “eco”. The product name n2 and the product name n3 include a token t ₁₁ “brandA”. The product name n4 includes tokens t ₁₁ “brandA” and t ₁₃ “eco”.

  In this way, the number of product names used for processing can be limited by setting the product name including the token of the product page title included in the product page to be processed, so the processing load can be reduced. It becomes.

  The estimation unit 13 is a part that estimates a feature related to the number of tokens included in the product page title of the product page to be processed. Specifically, the estimation unit 13 determines the processing target according to the characteristics of one or more product pages (not limited to the processing target product page) input by a store that handles the products related to the processing target product page. The characteristic regarding the token number of the product page title of the product page is estimated.

  As described above, the product page acquisition unit 11 can acquire a product page (including a product page other than the processing target product page and the processing target product page) input by the store that handles the product related to the processing target product page. Therefore, the estimation unit 13 acquires the product page used for the estimation process via the product page acquisition unit 11.

  The characteristic regarding the number of tokens can indicate, for example, that the number of tokens is less than a predetermined level, or that the number of tokens is equal to or higher than a predetermined level. Below, the estimation process by the estimation part 13 is demonstrated in detail.

  The estimation unit 13 estimates the characteristics related to the token number of the product page title of the product page to be processed based on the data amount constituting the product page input by the store handling the product related to the product page to be processed. Also good.

  The characteristic regarding the token number of the product page title of the product page to be processed is reflected in the data amount constituting the product page input by the store handling the product related to the product page to be processed. That is, as the amount of data constituting a plurality of product pages input by a store increases, the number of tokens of the product page title of the product page of the store tends to increase.

  In view of such a trend, the characteristics related to the number of tokens of the product page title of the product page to be processed are estimated based on the data amount constituting the product page input by the store handling the product related to the product page to be processed. Thus, the reliability of the feature related to the estimated number of tokens is improved.

  More specifically, the estimation unit 13 determines the product page title of the product page to be processed when the data amount of the product page input by the store handling the product related to the product page to be processed is equal to or greater than a predetermined amount. When the number of tokens is estimated to be greater than or equal to a predetermined level and the amount of data is less than the predetermined amount, it is estimated that the number of tokens of the product page title of the product page to be processed is less than the predetermined level. In the comparison between the data amount and the predetermined amount, a statistical value such as an average value or a total value of the data amounts of a plurality of product pages may be used.

  In addition, the estimation unit 13 estimates a feature related to the number of tokens of the product page title of the product page to be processed based on the number of words of the text that configures the product page input by the store that handles the product related to the product page to be processed. It is good to do. That is, in the above estimation process, the data amount constituting the product page input by the store handling the product related to the product page to be processed is used, but the number of text words constituting the product page is used as this data amount. It is good.

  As the number of texts constituting the plurality of product pages input by the store increases, the number of tokens of the product page title of the product page of the store tends to increase. In view of such a tendency, based on the number of text words constituting the product page input by the store handling the product related to the product page to be processed, there is a feature regarding the token number of the product page title of the product page to be processed. Since it is estimated, the reliability of the estimated feature is improved.

  More specifically, when the number of texts constituting the product page input by the store handling the product related to the product page to be processed is equal to or greater than a predetermined amount, the estimation unit 13 determines the product on the product page to be processed. If the number of tokens in the page title is estimated to be greater than or equal to a predetermined level and the number of words is less than a predetermined amount, it is estimated that the number of tokens in the product page title of the product page to be processed is less than a predetermined level. In the comparison between the number of words constituting the product page and a predetermined amount, a statistical value such as an average value or a total value of the number of words of a plurality of product pages may be used.

  In addition, the estimation unit 13 relates to the number of tokens of the product page title of the product page to be processed based on the number of tokens of the product page title included in the product page input by the store handling the product related to the product page to be processed. The feature may be estimated. That is, in the estimation process, the data amount or the number of words constituting the product page input by the store handling the product related to the product page to be processed is used, but instead of these, the product page title of the product page included in the product page is used. The number of tokens may be used.

  As the number of product page title tokens included in a plurality of product pages input by a store increases, the number of product page title tokens of the product page to be processed in the store tends to increase. In view of such a trend, the number of tokens of the product page title of the product page to be processed based on the number of tokens of the product page title included in the product page input by the store handling the product related to the product page to be processed Therefore, the reliability of the estimated feature is improved.

  More specifically, when the number of tokens of the product page title included in the product page input by the store that handles the product related to the product page to be processed is equal to or greater than a predetermined amount, the estimation unit 13 performs the product to be processed. If the number of tokens in the product page title of the page is estimated to be greater than or equal to a predetermined level and the number of tokens is less than the predetermined amount, the number of tokens in the product page title of the product page to be processed is less than the predetermined level Estimated. Note that in the comparison between the number of product page title tokens included in the product page and a predetermined amount, a statistical value such as an average value or total value of the number of product page title tokens in a plurality of product pages may be used.

  In addition, when the ratio of the number of tokens of the product page title included in the product page to be processed with respect to the number of tokens included in the product name acquired by the product name acquisition unit 12 is greater than or equal to a predetermined ratio, the estimation unit 13 If the number of tokens in the product page title of the processing target product page is estimated to be greater than the predetermined level and the ratio is less than the predetermined level, the number of tokens in the product page title of the processing target product page is less than the predetermined level It is good also as estimating. When there are a plurality of product names acquired by the product name acquisition unit 12, for example, a product name having the maximum number of tokens may be used for estimation among the plurality of product names.

  The index value such as the ratio of the number of tokens of the product page title included in the processing product page to the number of tokens included in the product name directly reflects the characteristics related to the number of tokens of the product page title of the processing product page. The Since the feature regarding the number of tokens is estimated based on the index value, the reliability of the estimated feature can be improved.

  The execution unit 14 executes the first process when the feature related to the number of tokens included in the product page title of the product page estimated by the estimation unit 13 indicates that the number of tokens is less than a predetermined level. When the feature indicates that the number of tokens is equal to or greater than a predetermined level, the second processing is executed.

  The first process is a process in which, when the number of tokens included in the product page title is small, a score that appropriately reflects the identity between the product indicated by the product name and the product related to the product page is calculated. On the other hand, the second process is a process in which, when the number of tokens included in the product page title is large, a score that appropriately reflects the identity between the product indicated by the product name and the product related to the product page is calculated. . Hereinafter, the first and second processes will be described in detail.

  The first process is a score indicating the identity between the product indicated by the product name and the product related to the product page, with the ratio of the other of the token included in the product name and the token included in the product page title being matched. Is calculated as follows.

The first process may be, for example, a process of calculating a Jaccard coefficient of a set of tokens included in the product name and a set of tokens included in the product page title of the product page to be processed as a score. When the set of tokens included in the product name is X and the set of tokens included in the product page title is Y, the Jaccard coefficient J (X, Y) of the set X and the set Y is expressed as follows.
J (X, Y) = | X∩Y | / | X∪Y |

  In the present embodiment, the Jaccard coefficient is calculated as the first score, but the index value indicating the ratio of the other match to one of the token included in the product name and the token included in the product page title. If so, the first score is not limited to the Jaccard coefficient.

  On the other hand, in the second process, the ratio of one of the token included in the product name and the token included in the product page title is included in the other, and the identity between the product indicated by the product name and the product related to the product page is determined. It is a process which calculates as a score to show.

  In the second process, for example, the token included in the product name and the token included in the product page title are shared with respect to the smaller number of tokens included in the product name and the token number included in the product page title. It may be a process of calculating the ratio of the number of tokens that have occurred as a score.

Below, the example of the score calculation by the 1st and 2nd process in the execution part 14 is demonstrated. 7, calculates a product page title T ₁ of the product page to be processed shown in FIG. 5, the respective trade names n1~n4 shown in FIG. 6, according to the first and second processing of scores indicating the identity Results are shown. In FIG. 7, the score A is a score calculated by the first process, and the score B is a score calculated by the second process.

The product page title T ₁ and names n1, specifically described the score calculation by the first and second processing. First, calculation of the score A by the first process will be described.

The execution unit 14 divides the product page title T ₁ “brandA abc08-bk eco air cleaner” into five tokens. The divided five tokens are a token t ₁₁ “brandA”, a token t ₁₂ “abc08-bk”, a token t ₁₃ “eco”, a token t ₁₄ “air”, and a token t ₁₅ “purifier”. On the other hand, the product name n1 “brandA abc08-bk eco” includes three tokens (“brandA”, “abc08-bk”, “eco”).

A set of tokens of the commodity page title T ₁ and X, and the set of tokens of trade names n1 and Y,
| X∩Y | = 3
| X∪Y | = 5
Since a, score A indicating the identity of the commodity page title T ₁ and the product name n1 is,
J (X, Y) = 0.6 (= 3/5)
It becomes.

Subsequently, the calculation of the score B by the second process will be described. In token of goods page title T ₁ of the token and product names n1, the number of tokens to the co-occurrence is 3. In addition, of the number of tokens products page title T ₁ of the token number and product names n1, lesser number of tokens is 3. Therefore, the score B indicating the identity of the commodity page title _{T 1} and names n1 is 1 (= 3/3). For even scores A and scores B indicates the identity of the respective commodity page title T ₁ and names N2 to N4, as illustrated in FIG. 7, is calculated in the same manner.

When FIG. 7 is referred, the score A regarding the brand name n1 has shown the value remarkably large compared with the score A regarding the brand names n2-n4. On the other hand, none of the scores B relating to the product names n1 to n4 show a significantly large value. Thus, in the commodity page title T ₁ number of tokens is relatively small contained identity determined with each trade name, from the score B calculated by the second processing, the score A calculated by the first processing Is preferred.

Next, calculation of a score indicating the identity between a product page title and a product name having a relatively large number of tokens will be described. Figure 8 shows a product page title T ₂ containing a number of tokens from the commodity page title T ₁ shown in FIG. Product page title T ₂ can be divided into 18 tokens. 9, calculates a product page title T ₂ of the product page to be processed shown in FIG. 8, with the respective trade names n1~n4 shown in FIG. 6, according to the first and second processing of scores indicating the identity Results are shown. In FIG. 9, the score A is a score calculated by the first process, and the score B is a score calculated by the second process.

The product page title T ₂ and names n1, specifically described the score calculation by the first and second processing. First, calculation of the score A by the first process will be described.

The execution unit 14 divides the product page title T ₂ “brandA abc08-bk eco air purifier humidification function pollen dust black free shipping cash on delivery impossible card OK energy saving 30 m ² ” into 18 tokens. The divided 18 tokens are token t ₂₀₁ “brandA”, token t ₂₀₂ “abc08-bk”, token t ₂₀₃ “eco”, token t ₂₀₄ “air”, token t ₂₀₅ “cleaner”, token t ₂₀₆ “ Humidification ”, token t ₂₀₇ “ function ”, token t ₂₀₈ “ pollen ”, token t ₂₀₉ “ dust ”, token t ₂₁₀ “ black ”, token t ₂₁₁ “ shipping ”, token t ₂₁₂ “ free ”, token t ₂₁₃ “ _price ” Pull ”, token t ₂₁₄ “ impossible ”, token t ₂₁₅ “ card ”, token t ₂₁₆ “ OK ”, token t ₂₁₇ “ energy saving ”, token t ₂₁₈ “ 30 m ² ”. On the other hand, the product name n1 “brandA abc08-bk eco” includes three tokens (“brandA”, “abc08-bk”, “eco”).

A set of tokens of the commodity page title T ₂ and X, and the set of tokens of trade names n1 and Y,
| X∩Y | = 3
| X∪Y | = 18
Since a, score A indicating the identity of the commodity page title T ₂ and the product name n1 is,
J (X, Y) = 0.167 (= 3/18)
It becomes.

Next, explaining the calculation of the score B according to the second processing on the product page title T _2. In token of the product page title T ₂ of the token and product names n1, the number of tokens to the co-occurrence is 3. In addition, of the number of token items of token page title T ₂ and product names n1, lesser number of tokens is 3. Therefore, the score B indicating the identity of the commodity page title _{T 2} and names n1 is 1 (= 3/3). For even scores A and scores B indicates the identity of the respective commodity page title T ₁ and names N2 to N4, as illustrated in FIG. 9, is calculated in the same manner.

Referring to FIG. 9, none of the scores A related to the product names n1 to n4 shows a significantly large value. On the other hand, the score B related to the product name n1 is significantly larger than the score B related to the product names n2 to n4. Therefore, in determining the identity between the product page title T ₂ and the product names that contain a relatively large number of tokens, the score B calculated by the second process is calculated from the score A calculated by the first process. Is preferred.

In estimating Product Page token number more or less of the features of the title, by setting the predetermined value as a reference to an appropriate value, for example, the token number is relatively small items page title such as Product Page Title T ₁ is estimated to have a characteristic number of tokens is smaller, also the token number is relatively large product page titles like products page title T ₂ are, are estimated to have a number of features many tokens.

As such by the features relates token number is estimated, the identity of the commodity page title T ₁ and each product name is determined based on the score calculated by the first processing, commodity page title T ₂ And the identity of each product name are determined based on the score calculated by the second process. Note that the predetermined value serving as a reference in estimating the characteristics of the number of tokens is, for example, a value set by design.

  Therefore, according to the product page to be processed, it is possible to calculate a score that appropriately reflects the identity, and a score that appropriately reflects the identity is used. It is possible to specify a product name having the same identity.

  Referring again to FIG. 1, the specifying unit 15 is a part that specifies a product name having the same identity between the product page and the product according to the score calculated by the first process or the second process.

For example, when a specific product name identity with the product page containing product page title T _1, since the score by the first process is calculated, the specifying unit 15, according to the score A shown in FIG. 7 The product name n1 is specified as the product name having the identity.

Further, for example, upon the particular trade name identity with the product page containing product page title T ₂ are, since the score by the second process is calculated, the specifying unit 15, the score B shown in FIG. 9 Accordingly, the product name n1 is specified as the product name having the same identity.

  The specifying unit 15 may store the official product name specified in this way in the product page storage unit 21 in association with the product page. Further, the specifying unit 15 may generate association data that forms a group of a plurality of product pages having the same product name specified as having the same identity.

  Further, the product identification device 1 monitors the update status of the product page title of the product page in the product page storage unit 21 by the store, and when the degree of change of the product page title is within a predetermined range, In the process and the second process, the score is calculated by the process used before, and when the degree of change of the product page title exceeds a predetermined range, the estimation unit 13 performs the characteristic estimation process regarding the number of tokens, The score may be calculated by one of the first process and the second process according to the estimated feature.

  It should be noted that the product identification device 1 has a frequency or rate at which a score is calculated by the second process as compared to the first process for the process of identifying a product name that is identical to the product page input by a certain store. In the case where it is equal to or greater than a predetermined level, it may further include a notification unit that notifies the store of a request for suppression of including words other than the official product name (for example, so-called spam words) in the product page. .

  Next, with reference to FIG. 10, operation | movement of the goods specific device 1 of this embodiment is demonstrated. FIG. 10 is a flowchart showing the processing content of the product specifying method implemented in the product specifying device 1.

  First, the product page acquisition unit 11 acquires a product page to be processed (S1). Next, the product name acquisition unit 12 acquires the official product name stored in the product name storage unit 22 (S2).

  The estimation unit 13 estimates a feature related to the number of tokens included in the product page title of the product page to be processed (S3). Subsequently, the execution unit 14 determines whether or not the feature related to the number of tokens included in the product page title of the product page estimated in Step S3 indicates that the number of tokens is less than a predetermined level (S4). If it is determined that the characteristic regarding the number of tokens indicates that the number of tokens is less than a predetermined level, the process proceeds to step S5. On the other hand, if it is not determined that the feature related to the number of tokens indicates that the number of tokens is less than a predetermined level, the process proceeds to step S6.

  In step S5, the execution unit 14 calculates a score indicating the identity between the product indicated by the product name and the product related to the product page by the first processing (S5). On the other hand, in step S6, the execution unit 14 calculates a score indicating the identity between the product indicated by the product name and the product related to the product page by the second process (S6).

  And the specific | specification part 15 specifies the goods name which has the identity of a goods page and goods according to the score calculated by the 1st process or the 2nd process (S7).

  Next, a product specifying program for causing a computer to function as the product specifying device 1 will be described with reference to FIG. The product specifying program p1 includes a main module m10, a product page acquisition module m11, a product name acquisition module m12, an estimation module m13, an execution module m14, and a specification module m15.

  The main module m10 is a part that comprehensively controls the product specifying process. The functions realized by executing the product page acquisition module m11, the product name acquisition module m12, the estimation module m13, the execution module m14, and the specific module m15 are each a product page acquisition unit 11 of the product specification device 1 shown in FIG. The functions of the product name acquisition unit 12, the estimation unit 13, the execution unit 14, and the specification unit 15 are the same.

  The product specifying program p1 is provided by a storage medium d1 such as a CD-ROM, a DVD-ROM, or a semiconductor memory, for example. Further, the product identification program p1 may be provided via a communication network as a computer data signal superimposed on a carrier wave.

  According to the product specifying device 1, the product specifying method, and the product specifying program p1 of the present embodiment described above, the feature regarding the number of tokens included in the product page title of the product page is estimated, and according to the estimated feature, By the process 1 or the second process, a process of calculating a score indicating the identity between the product indicated by the product name and the product related to the product page is executed. The first process is a process in which, when the number of tokens included in the product page title is small, a score that appropriately reflects the identity between the product indicated by the product name and the product related to the product page is calculated. On the other hand, the second process is a process in which, when the number of tokens included in the product page title is large, a score that appropriately reflects the identity between the product indicated by the product name and the product related to the product page is calculated. . The first process is performed when the characteristic regarding the estimated number of tokens indicates that the number of tokens is less than a predetermined level, and the second process is performed when the number of tokens indicates that the number of tokens is equal to or greater than the predetermined level. Is executed, a score that appropriately reflects the identity is calculated according to the product page to be processed. Since the score appropriately reflecting the identity is used, it is possible to specify the product name having the same identity as the product page with high accuracy.

  The present invention has been described in detail based on the embodiments. However, the present invention is not limited to the above embodiment. The present invention can be variously modified without departing from the gist thereof.

DESCRIPTION OF SYMBOLS 1 ... Product specification apparatus, 11 ... Product page acquisition part, 12 ... Product name acquisition part, 13 ... Estimation part, 14 ... Execution part, 15 ... Specification part, 21 ... Product page storage part, 22 ... Product name storage part, d1 ... storage medium, m10 ... main module, m11 ... product page acquisition module, m12 ... product name acquisition module, m13 ... estimation module, m14 ... execution module, m15 ... specific module, p1 ... product specific program.

Claims (10)

A product page acquisition means for acquiring a product page to be processed, wherein the product page includes a text representing the product related to the product page and includes a product page title that can be divided into tokens composed of one or more characters. , Product page acquisition means,
Product name acquisition means for acquiring at least one preset product name, wherein the product name includes product name acquisition means including one or more tokens composed of one or more characters;
Estimating means for estimating characteristics relating to the number of tokens included in the product page title of the product page;
When the feature estimated by the estimating means indicates that the number of tokens is less than a predetermined level, a first process is executed, and the feature indicates that the number of tokens is equal to or higher than a predetermined level. In this case, execution means for executing a second process, wherein the first process determines a ratio of the other of the token included in the product name and the token included in the product page title to the other. , A score indicating the identity between the product indicated by the product name and the product related to the product page is calculated, and the second process includes a token included in the product name and a token included in the product page title. An execution means for calculating a ratio in which one of the two is included in the other as a score indicating the identity;
A specifying means for specifying the product name having the same identity as the product page according to the score calculated by the first process or the second process;
A product identification device comprising: The product name acquisition means acquires a product name set in advance by a person other than the store handling the product related to the processing target product page,
The estimation means estimates a feature related to the number of tokens of the product page title of the product page to be processed according to the feature of one or more product pages input by the store.
The product identification device according to claim 1. The estimation means estimates a feature related to the token number of the product page title of the product page to be processed based on a data amount constituting the product page input by a store handling the product related to the product page to be processed. ,
The product identification device according to claim 2. The estimation means estimates that the number of tokens of the product page title of the product page to be processed is greater than or equal to a predetermined level when the data amount is greater than or equal to a predetermined amount, and if the data amount is less than the predetermined amount , Estimate that the number of tokens in the product page title of the product page being processed is less than a certain amount,
The product identification device according to claim 3. The estimation means estimates a feature related to the number of tokens of the product page title of the product page to be processed based on the number of texts constituting the product page input by the store handling the product related to the product page to be processed. ,
The product identification device according to claim 3 or 4. The estimation means is characterized by the number of tokens of the product page title of the product page to be processed based on the number of tokens of the product page title included in the product page input by the store handling the product related to the product page to be processed. Estimate
The product identification device according to claim 3 or 4. The estimation means performs processing when the ratio of the number of tokens of the product page title included in the product page to be processed with respect to the number of tokens included in the product name acquired by the product name acquisition means is equal to or greater than a predetermined ratio. When the number of tokens of the product page title of the target product page is estimated to be greater than a predetermined level, and the ratio is less than the predetermined level, the number of tokens of the product page title of the target product page is less than the predetermined level To estimate,
The product identification device according to claim 2. The product name acquisition means acquires a product name including at least one product page title token included in the product page to be processed.
The product identification device according to any one of claims 1 to 7. A product identification method in a product identification device,
A product page acquisition step for acquiring a product page to be processed, wherein the product page includes a product page title that includes text representing the product related to the product page and can be divided into tokens including one or more characters. , Product page acquisition step,
A product name acquisition step of acquiring at least one preset product name, wherein the product name includes one or more tokens including one or more characters;
An estimation step of estimating a feature related to the number of tokens included in the product page title of the product page;
If the feature estimated in the estimating step indicates that the number of tokens is less than a predetermined level, a first process is executed, and the feature indicates that the number of tokens is equal to or higher than a predetermined level. In this case, it is an execution step of executing a second process, wherein the first process determines the ratio of the other match to one of the token included in the product name and the token included in the product page title. , A score indicating the identity between the product indicated by the product name and the product related to the product page is calculated, and the second process includes a token included in the product name and a token included in the product page title. An execution step of calculating a ratio in which one of the two is included in the other as a score indicating the identity;
In accordance with the score calculated by the first process or the second process, a specifying step of specifying the product name having the same product as the product page;
A product identification method. On the computer,
A product page acquisition function for acquiring a product page to be processed, wherein the product page includes a product page title that includes text representing the product related to the product page and can be divided into tokens including one or more characters. , Product page acquisition function,
A product name acquisition function for acquiring at least one preset product name, wherein the product name includes a product name acquisition function including one or more tokens including one or more characters;
An estimation function for estimating a characteristic regarding the number of tokens included in the product page title of the product page;
When the feature estimated by the estimation function indicates that the number of tokens is less than a predetermined level, a first process is executed, and the feature indicates that the number of tokens is equal to or higher than a predetermined level. In this case, the execution function executes a second process, and the first process determines a matching ratio of the other of the token included in the product name and the token included in the product page title. , A score indicating the identity between the product indicated by the product name and the product related to the product page is calculated, and the second process includes a token included in the product name and a token included in the product page title. An execution function for calculating a ratio indicating that one of the two is included in the other as a score indicating the identity;
According to the score calculated by the first process or the second process, a specifying function for specifying the product name having the same product as the product page;
Product identification program that realizes

Images