JP5321777B2 - Product search device and product search method having function of presenting reference keyword - Google Patents

Description

  The present invention relates to a technology for searching for product data that matches a search keyword input by a user from a database, and more particularly to a technology for dealing with an input error of a search keyword.

  For shopping sites and actual stores established on the Internet (for example, book stores, music software stores, home appliance mass retailers, etc.), the products that the user desires from the products handled at the shopping sites and actual stores Must be able to search quickly.

  However, users rarely remember the spelling of words related to products, such as the name of the product, and when searching for products using the name of the product as a search keyword, Since there are cases where the search fails, it is desirable that when searching for products, it is possible to present to the user words that are estimated from the search keywords entered by the user.

As an invention for presenting a keyword other than the search keyword input by the user to the user, for example, in Patent Document 1, a synonym or synonym corresponding to the search keyword input by the user is provided with a synonym / synonym dictionary. An information search system that extracts word / synonym information is disclosed.
JP 2003-216634 A

  However, if the invention disclosed in Patent Document 1 is used, a synonym or synonym corresponding to the search keyword input by the user can be presented to the user. However, the invention disclosed in Patent Document 1 Is not an invention for dealing with a search keyword input error.

  Therefore, the present invention can search for a product that is inferred from the search keyword input by the user when there is an input error of the search keyword when searching the database for product information that matches the search keyword input by the user. An object of the present invention is to provide an apparatus and a method capable of providing a word to a user.

A first invention for solving the above-described problem is a product search method for searching for product data that matches a search keyword input by a user, in which product data including a product related word that is a word related to the product is stored. When the device provided with the product database receives a search request for the product data from the terminal device operated by the user, the device includes the search keyword transmitted from the terminal device as the product-related word. Step a for retrieving data from the product database, and when the device determines that the product data search has failed in step a, the device stores the product related words and the search stored in the product database. the number of the similarity between the keywords, the first sub-string obtained from the search keyword included in the product related word The value obtained by dividing the total number of the first partial character strings and the value obtained by dividing the number of the second partial character strings obtained from the product related words included in the search keyword by the total number of the second partial character strings. calculated by multiplying the bets, step b extracting the items related words meet the criterion similarity is defined by reference keyword, the device, the terminal device the reference keywords extracted in step b A product search method comprising: generating display data to be displayed on the terminal device and transmitting the display data to the terminal device.

2nd invention is the goods search method as described in 1st invention, Comprising : The said goods related word is produced | generated using the word contained in the specification or description of goods memorize | stored in the said goods database. It is characterized by that .

A third invention is a commodity search method according to the first or second invention, wherein in the step b, the device uses partial character strings having different numbers of characters as the partial character strings. And

4th invention is the goods search method as described in any one of 1st invention from 3rd invention, Comprising: In the said step b, the said apparatus contains the said search keyword as the said product related word. When the number of product data is 0 or less than a predetermined number, the device determines that the device has failed in the search for the product data.

5th invention is the goods search method as described in any one of 1st invention from 4th invention, Comprising: The said apparatus is equipped with the said goods database for every category of the said goods, The said step a is When the product data search request is received from the terminal device operated by the user, the device includes the product data including the search keyword transmitted from the terminal device as the product related word, from the terminal device. In the step of searching from the product database corresponding to the category of the transmitted product, when the device determines that the search of the product data has failed in the step a, the device the product-related words stored in the product database corresponding to the transmitted said category of items from the device and said search keyword Calculates the degree of similarity, characterized in that it is a step of extracting the product related words meet the criterion similarity is defined by reference keyword.

Further, a sixth invention is a product search device for searching for product data that matches a search keyword input by a user, a product database storing product data including product related words that are words related to the product, A product search means for searching the product database for the product data including the search keyword transmitted from the terminal device as the product related word when receiving a search request for the product data from a terminal device operated by a user; When the product search unit fails to search the product data, the similarity between the product related word stored in the product database and the search keyword is obtained as a first partial character string obtained from the search keyword. A value obtained by dividing the number included in the product-related word by the total number of the first partial character strings, and a value obtained from the product-related word. Substring is calculated by multiplying the value of the number obtained by dividing by the total number of the second partial character string included in the search keyword, the product related words meet the criterion similarity is defined A reference keyword extracting means for extracting the reference keyword as a reference keyword, display data for displaying the reference keyword extracted by the reference keyword extraction means on the terminal device, and transmitting the generated display data to the terminal device And a display means.

A seventh invention is the product search device according to the sixth invention, wherein the product-related word is generated by using a word included in a product specification or an explanatory text stored in the product database. and wherein the der Rukoto thing was.

Eighth invention is the product search device according to the sixth invention or the seventh, the reference keyword extracting means, as the partial string, and characterized to have access to a substring of the different characters To do.

9th invention is the goods search apparatus as described in any one of 6th invention to 8th invention , Comprising : The said goods search means of the said product data containing the said search keyword as said goods related word When the number of cases is 0 or less than a predetermined number, it is determined that the product data search has failed.

A tenth invention is the product search device according to any one of the sixth invention to the ninth invention, wherein the product search device is a web server arranged on the Internet, and the reference keyword display The means generates HTML data for displaying the reference keyword in the form of a hyperlink in which a URL of the product search device for operating the product search means is embedded as the display data.

An eleventh invention is the product search device according to any one of the sixth invention to the tenth invention, wherein the product search device includes the product database for each category of the product, and the product When the search unit receives a search request for the product data from the terminal device operated by the user, the search unit transmits the product data including the search keyword transmitted from the terminal device as the product-related word from the terminal device. In the means for searching from the product database corresponding to the category of the product that has been made, the reference keyword extracting means, when the product search means fails to search for the product data, of the product sent from the terminal device calculating the product-related words and similarity between the search keywords stored in the item database corresponding to the category, class Degrees, characterized in that it is a means for extracting reference keyword matching the items related terms the criteria defined.

  According to the above-described invention, when the search keyword input by the user to the terminal device is used and the search for the product data fails, the product related word that meets the criteria for which the similarity to the search keyword is determined is used as the reference keyword. By extracting and displaying the extracted reference keywords on the terminal device, it is possible to present to the user product-related words that may be correct with respect to the input search keyword, and input due to lack of user knowledge and carelessness Can make up for mistakes.

  The more product data stored in the product search device, the greater the number of product related words stored in the product search device. Therefore, the similarity between the search keyword and the product related word can be calculated efficiently. In order to efficiently calculate the similarity between the search keyword and the product related word, the edit distance between the product related word and the search keyword and the partial character string obtained from the search keyword are used as the product related word. By calculating the similarity between the search keyword and the product related word using the number included, the calculation amount of the similarity can be reduced.

  Furthermore, by setting the judgment criterion for input mistakes to not more than 0 and a predetermined number or less, even if product data corresponding to a search keyword with an input error is searched, product related terms that may be correct are displayed to the user. Can be presented.

  When the product search device according to the present invention is realized by a web server, when the reference keyword is displayed in the form of a hyperlink in which the URL of the product search device that operates the product search means is embedded, the user can use the reference keyword. It becomes possible to save the trouble of inputting.

  From here, the goods search apparatus 1 and the goods search method which concern on this invention are demonstrated in detail, referring a figure. FIG. 1 is a diagram illustrating an embodiment of a product search device 1 according to the present invention. In the embodiment illustrated in FIG. 1, the product search device 1 according to the present invention is realized by a web server installed on a network 4, and a client 2 serving as a terminal device operated by a user 3 is also connected to the network 4. Yes.

  A commodity search apparatus 1 according to the present invention includes a central processing unit 10 (CPU: Central Processing Unit), a random access memory 11 (RAM: Random Access Memory), a large-capacity storage device 13 (for example, a hard disk), and a network interface 12. Similarly, the client 2 includes a CPU 20, a random access memory 21, a large-capacity storage device 23 (for example, a hard disk), and a network interface 22.

  FIG. 2 is a functional block diagram of the product search device 1 and the client 2. As shown in FIG. 2, the client 2 interprets HTML data described in HTML (HyperText Markup Language) as a computer program for operating the CPU 20 of the client 2, and displays characters and images. A browser 200 to be displayed on the display of the client 2 is provided, and a user 3 of the client 2 accesses the product search device 1 using the browser 200 provided in the client 2.

  As shown in FIG. 2, the product search device 1 includes a word obtained from the product specification / description in association with product data such as product identification data, specifications, description, and images as a database. A product database 110 (DB: DataBase) in which various product related words used for searching for products are stored, and a product related word DB 111 serving as a dictionary of product related words, as means for realizing the present invention, Search the product DB 110 for product data that matches the search keyword transmitted from the client 2 as a means for executing the processing specific to the web server unit 100 that processes the access request from the browser 200 of the client 2 and the product search device 1. Product search means 101 to search for a product that matches the search keyword received from the client 2 From the product-related words stored in the product-related word DB 111, and a reference keyword extracting means 102 for extracting a high product related words of search keywords similarity received from the client 2 as a reference keyword.

  Each means provided in the product search device 1 is realized by a computer program for causing the CPU 10 of the product search device 1 to function as each means, and these computer programs are stored in the storage device 13 of the product search device 1. Has been.

  In the product DB 110 provided in the product search device 1, product data related to products is stored for each product, and each product data includes product identification data, specifications, explanations, images, and other data. A plurality of product related words, which are words obtained from the specifications and explanations, are included.

  Product-related words, which are words obtained from product specifications and explanations, etc., shall be applied to the noun phrase extraction rules determined by decomposing words and phrases contained in product specifications and explanations into parts of speech by morphological analysis. Is generated. The defined noun phrase extraction rule means concatenation of adjacent nouns, concatenation of adjacent unknown words, etc. For example, from “face”, “detection” and “function”, “face”, “detection” And product-related words such as “face”, “face detection”, “detection function”, and “face detection function” are generated in addition to the product-related words of “function”.

  The product search device 1 may be provided with a function for generating product related words from product specifications or explanations by morphological analysis, or the product search device 1 may use product related words generated by other devices. It may be used.

  The product related word DB 111 provided in the product search device 1 stores the product related words stored in the product DB 110 so that there is no duplication of product related words.

  The means that operates when the product search device 1 is accessed from the browser 200 of the client 2 is the web server means 100, and the web server means 100 has the function of the reference keyword display means according to the present invention.

  The web server means 100 of the product search device 1 uses the HTML data for allowing the user 3 to input a search keyword, acquires the search keyword from the client 2, and succeeds in searching for product data using the search keyword. Generates HTML data for displaying the search result on the browser 200 of the client 2, and generates HTML data for displaying the reference keyword of the search keyword when the search for the product data using the search keyword fails.

  When the product search apparatus 1 is accessed from the browser 200 of the client 2, the web server means 100 of the product search apparatus 1 transmits HTML data for allowing the user 3 to input a search keyword to the client 2. The HTML data defines a tag that defines a form for allowing the user 3 to input a search keyword, and a button for searching for product data that matches the search keyword input in the form and displaying the search result. At least tags to be described are described.

  FIG. 3 is a diagram for explaining a web page 5 including a form for inputting a search keyword. The HTML data for allowing the user 3 to input a search keyword is interpreted by the browser 200, so that the web page 5 makes a search request to the product search apparatus 1 and the form 50 for the user 3 to input the search keyword. A search button 51 is displayed.

  In the search button 51, an action tag indicating the transmission destination of the search keyword input in the form 50 is described. Here, the transmission destination of the search keyword described in the action tag is the product search means of the product search device 1. A URL to be called when executing 101 is described.

  When the user 3 inputs a search keyword into the form 50 and clicks the search button 51 with a pointing device such as a mouse, the search keyword input into the form 50 is transmitted to the product search device 1 and the product of the product search device 1 The search unit 101 operates, and the product search unit 101 searches the product DB 110 for product data including the search keyword delivered from the client 2 as a product related word.

  When the product search unit 101 can search for product data including the search keyword delivered from the client 2 as a product-related word, the product search unit 101 delivers the searched product data to the web server unit 100, and the web server unit 100 is delivered. HTML data for displaying a list of the merchandise data is generated.

  The HTML data generated by the web server means 100 is transmitted to the client 2 via the network 4, and the product search result is displayed on the display of the client 2 by the browser 200 of the client 2.

  When the product search unit 101 cannot search for product data including the search keyword delivered from the client 2 as a product related word, that is, when the search result is 0, the product search unit 101 extracts the search keyword as a reference keyword. Deliver to the means 102 and delegate the reference keyword extraction means 102 to extract the reference keyword of the search keyword delivered from the client 2.

  The reference keyword extraction unit 102 calculates the similarity between each of the product related words stored in the product related word DB 111 and the search keyword, and selects the product related words whose similarity meets a predetermined criterion. Extract as a reference keyword.

  Here, the criteria for extracting as a reference keyword based on the similarity may be arbitrary. For example, product-related words having a similarity equal to or higher than a certain threshold may be extracted as a reference keyword, You may extract a product related word as a reference keyword.

  Any method may be used to calculate the similarity between the product related word stored in the product related word DB 111 and the search keyword. As a method for calculating the similarity between the product related word and the search keyword, The first similarity calculation method for calculating the similarity by calculating the edit distance between the search keyword and the second similarity calculation method for calculating the similarity from the appearance frequency of the partial character string When selected and used, the similarity between the product-related word and the search keyword can be calculated efficiently.

  First, the first similarity calculation method for calculating the edit distance will be described. In the first similarity calculation method, the edit distance between the search keyword acquired from the client 2 and the product related word stored in the product related word DB 111 is calculated using a predetermined edit distance algorithm, and the calculation is performed. The reciprocal of the edited distance is used as the similarity.

  Generally, when calculating an edit distance between two character strings, an edit operation and a distance corresponding to the edit operation are defined. For example, the distance corresponding to deletion of one character is “1”, the distance corresponding to insertion of one character is “1”, and the distance corresponding to replacement of one character is “1”. The edit distance between two character strings is the minimum number of editing operations for converting one character string into the other character string.

  There are various editing distance algorithms for calculating the editing distance between two character strings by a computer. Here, a method using dynamic programming which is a known technique will be described. A character string S (length) that stores characters included in the search keyword, where “m” is the length of the search keyword for which the product search has failed and “n” is the length of the product-related word for calculating the similarity. m), a character string T (length n) for storing characters included in the product-related word for calculating the similarity, and a two-dimensional array D (m + 1, n + 1) are formed in the RAM 11 of the product search device 1, According to the recurrence formula shown in Equation 1, the value of each element of the two-dimensional array D (m + 1, n + 1) is calculated, and the value of the two-dimensional array D (m + 1, n + 1) is between the character string S and the character string T. Is calculated as the edit distance.

  From here, the above-mentioned content is demonstrated, demonstrating a specific example. FIG. 4 is a diagram illustrating a specific example of the first calculation method. In FIG. 4, the search keyword that failed to be searched is 7 characters “Laincup”, and the product-related word whose similarity is calculated is 7 characters “Linecup”.

  FIG. 4A is a diagram for explaining the two-dimensional array D (m + 1, n + 1) after the initial value is set according to Equation (3). As indicated by (3) in Equation 1, “0” is assigned to the two-dimensional array D (0, 0), and the i number “1-7” is assigned to the two-dimensional array D (i, 0). ”Are respectively substituted, and j numbers“ 1 to 7 ”are respectively substituted into the two-dimensional array D (0, j).

  FIG. 4B illustrates the two-dimensional array D (m + 1, n + 1) when i = 1 is completed. For the calculation of the two-dimensional array D (1, 1), the value in the dotted line frame is used, and “0” that is the value calculated from (1) and (2) of Equation 1 is substituted. Values calculated from (1) and (2) of Equation 1 are assigned to the two-dimensional array D (1, 2) to the two-dimensional array D (1, 7), respectively.

  FIG. 4C illustrates the two-dimensional array D (m + 1, n + 1) when i = 7 is completed. For the calculation of the two-dimensional array D (7, 7), the value in the dotted line frame is used, and “2” that is the value calculated from (1) and (2) of Equation 1 is substituted. The value of the two-dimensional array D (7, 7) is the edit distance between the search keyword “Laincup” and the product-related word “Linecup”, and the similarity between the search keyword “Laincup” and the product-related word “Linecup” is described above. As shown, it is the reciprocal of the edit distance, and its value is “0.5”.

  As described above, in the first similarity calculation method, the amount of calculation of the similarity depends on the two-dimensional array D (m + 1, n + 1) and can be simply calculated. It is possible to calculate efficiently.

  Next, a second similarity calculation method based on the partial character string will be described. In the second similarity calculation method, n consecutive rows (N-gram), preferably all 2-gram partial character strings, are extracted from each of the search keyword and the product related word that have failed to be searched, The similarity is calculated according to

  According to Equation 2, the value obtained by dividing the number of partial character strings S obtained from the search keyword contained in the product-related words by the total number of partial character strings S and the partial character string T obtained from the product-related words are used as the search keywords. The similarity between the search keyword and the merchandise related word is calculated by multiplying the number obtained by dividing the number included by the total number of partial character strings T.

  For example, an example in which the search keyword is “Laincup”, the product related word is “Linecup”, and the length of the partial character string is 2 characters will be described. The two-character partial character string S obtained from the search keyword “Laincup” includes six characters “La”, “ai”, “in”, “nc”, “cu”, and “up”, and the product-related word “Linecup”. The number of partial character strings S of the search keyword included in “” is three.

  In addition, the two-character partial character string T obtained from the product-related word “Linecup” is “Li”, “in”, “ne”, “ec”, “cu”, and “up”, and the search keyword The number of partial character strings T of search keywords included in “Laincup” is three.

  Therefore, the similarity between the search keyword “Laincup” and the product-related word “Linecup” is “0.250” according to Equation 2.

  In this way, if the number of characters in the partial character string is 2 in the second similarity calculation method, the number of partial character strings included in the search keyword is “number of characters in the search keyword−1”. Since the number of partial character strings included in the word is “number of characters of product-related words—1”, the amount of calculation for calculating the similarity between the product-related word and the search keyword is small.

  In the example described above, a 2-gram partial character string is used. However, N-gram is not limited to 2-gram, and a 3-gram or 4-gram partial character string may be used for similarity calculation. In addition, partial character strings obtained from a plurality of N-grams having different N values may be used for similarity calculation.

  For example, the similarity can be calculated according to Equation 2 using 2-gram and 3-gram partial character strings. When the 2-gram and 3-gram partial character strings are used for the similarity calculation, the partial character string S is the 2-gram and 3-gram partial character strings obtained from the search keyword, and the search keyword is “Laincup”. ”, The partial character string S includes“ La ”,“ ai ”,“ in ”,“ nc ”,“ cu ”,“ up ”,“ Lai ”,“ ain ”,“ inc ”,“ ncu ”and There are 11 "cups".

  The partial character string T is a 2-gram and 3-gram partial character string obtained from the product-related words. If the product-related word is “Linecup”, the partial character string T is “Li”, “in”. , “Ne”, “ec”, “cu”, “up”, “Lin”, “ine”, “nec”, “ecu”, and “cup”.

  Therefore, according to Equation 2, the similarity SIM (S, T) when using the 2-gram and 3-gram partial character strings is calculated as 4/11 × 4/11, and the value is approximately 0.132.

  Furthermore, the similarity can be calculated according to Equation 2 using 2-gram, 3-gram and 4-gram partial character strings. When the 2-gram, 3-gram, and 4-gram partial character strings are used for the similarity calculation, the partial character string S is the 2-gram, 3-gram, and 4-gram partial characters obtained from the search keyword. If the search keyword is “Laincup”, the partial character string S is “La”, “ai”, “in”, “nc”, “cu”, “up”, “Lai”, “ain”, “ain”, There are 15 "inc", "ncu", "cup", "Lain", "ainc", "incu", and "ncup".

  The partial character string T is a 2-character, 3-gram, and 4-gram partial character string obtained from the product-related word. If the product-related word is “Linecup”, the partial character string T is “Li”. , “In”, “ne”, “ec”, “cu”, “up”, “Lin”, “ine”, “nec”, “ecu”, “cup”, “Line”, “inec”, “ necu "and" ecup ".

  Therefore, according to Equation 2, the similarity SIM (S, T) when using the 2-gram, 3-gram, and 4-gram partial character strings is calculated as 4/15 × 4/15, The value will be about 0.071.

  If the similarity is calculated using partial character strings obtained from a plurality of N-grams with different N, the amount of calculation increases. However, since a plurality of N-grams with different N are used, the similarity There is an advantage that accuracy is improved.

  The reference keyword extraction unit 102 determines that the calculation of the similarity between the search keyword that failed in the search and the product-related word is performed for all the product-related words stored in the product-related word DB 111 with the contents as described above. Product-related terms that match the degree of similarity with the specified criteria are extracted as reference keywords.

  When the reference keyword extraction unit 102 extracts the reference keyword as a search keyword, the reference keyword extraction unit 102 delivers the extracted reference keyword to the web server unit 100, and the web server unit 100 searches for product data using the search keyword transmitted from the client 2. In addition to the message indicating the failure, HTML data is generated as display data for displaying the reference keyword delivered from the reference keyword extraction unit 102, and the generated HTML data is transmitted to the client 2.

  FIG. 5 is a diagram for explaining the web page 6 on which reference keywords are displayed. As shown in FIG. 3, in addition to the contents of the web page 5 shown in FIG. 3, a message 62 indicating that the search using the search keyword acquired from the client 2 has failed (here, 0 search results) Then, the reference keyword text 63 (here, Linecup) delivered from the reference keyword extraction unit 102 is displayed.

  When the reference keyword text 63 is displayed on the web page 6 of FIG. 5, the reference keyword is linked to the text 63, and when the hyperlink is clicked, the reference keyword is transmitted to the product search device 1. , It is possible to save the user from inputting a reference keyword.

  In the present embodiment, the product search device 1 generates HTML data as display data for displaying the reference keyword. However, the display data generated by the product search device 1 has a format other than HTML data. It may be. Naturally, the format of the display data generated by the product search device 1 depends on the specifications of the terminal device on which the display data is displayed.

  From here, the product search method according to the present invention will be described. FIG. 6 is a flowchart showing the procedure of the product search method according to the present invention. The product search method according to the present invention is a method executed in a device such as the product search device 1 described so far. First, the product search device 1 searches the product data from the client 2 that requested the search. A search keyword used when searching for product data is acquired (S1). The web server unit 100 of the commodity search apparatus 1 described above acquires a search keyword from the client 2 by using the web page 5 as illustrated in FIG.

  When the web server means 100 of the product search device 1 acquires the search keyword from the client 2, the search keyword is transferred to the product search means 101, and the product search means 101 acquires from the product DB 110 in which various product data are stored. The product data including the searched keyword as a product related word is searched (S2).

  The product search unit 101 of the product search device 1 branches the process according to the search result of the product data (S3). When the product data search is successful, the product search unit 101 delivers the search result of the product data to the web server unit 100. The means 100 generates HTML data for displaying the search result of the product data, transmits the generated HTML data to the client 2 (S4), and ends this procedure.

  When the product data search fails, the product search unit 101 of the product search device 1 delivers the search keyword acquired from the client 2 to the reference keyword extraction unit 102, and the reference keyword extraction unit 102 is stored in the product related term DB 111. The similarity between all the product related words and the search keyword is calculated according to the first similarity calculation method or the second similarity calculation method described above, and the product related words that match the defined criteria are referred to Extracted as a keyword (S5).

  The product search unit 101 of the product search device 1 delivers the extracted reference keyword to the web server unit 100, and the web server unit 100 of the product search device 1 displays HTML data that displays the reference keyword delivered from the reference keyword extraction unit 102. And the generated HTML data is transmitted to the client 2 so that the reference keyword extracted by the reference keyword extracting means 102 is displayed on the client 2 (S6).

  The present invention is not limited to the embodiments described so far, and various modifications and changes can be made. For example, the product search device 1 has been described as a single device so far, but the product search device 1 may be composed of a plurality of servers, and the function of the product search device 1 is one of shopping sites. It can also be provided as a function.

  Assuming that the product search device 1 is applied to a shopping site, product data is classified and managed for each product category in the shopping site, so the product DB 110 and product related terms provided in the product search device 1 are managed. The DB 111 is also preferably provided for each product category.

  When the product DB 110 and the product related term DB 111 are provided for each product category, the web server means 100 of the product search device 1 transmits HTML data for acquiring the product category before acquiring the search keyword from the client 2. The product category is acquired from the client 2.

  When the product category is acquired from the client 2, the product search unit 101 of the product search device 1 searches for product data from the product DB 110 corresponding to the product category acquired from the client 2, and the product search device The first reference keyword extraction unit 102 extracts reference reference keywords from the product related term DB 111 corresponding to the product category acquired from the client 2.

  As described above, the product search device 1 includes the product DB 110 and the product related word DB 111 for each product category, and selects and uses the product DB 110 and the product related word DB 111 corresponding to the product category acquired from the client 2. Since the product related words whose similarity is calculated with the search keyword are limited to the product related words stored in the product related word DB 111 corresponding to the product category acquired from the client 2, the search keyword and the product related The amount of calculation of the similarity with the word can be suppressed.

  Furthermore, although the goods search apparatus 1 which concerns on this invention was demonstrated as a web server in embodiment mentioned above, the goods search apparatus 1 which concerns on this invention is applicable also to the store terminal installed in a real store.

  When the product search device 1 according to the present invention is used as a store terminal, the terminal device is installed in an actual store, and the product search device 1 is realized by connecting to the terminal device via a LAN (LAN: Local Area Network). Alternatively, the terminal device and the product search device 1 are realized as an integrated device.

Further, in the above-described embodiment, the product search device 1 according to the present invention determines that the product data search has failed when the search result of the product data that matches the search keyword input by the user 3 is zero. The reference keyword was extracted, but when the search result of the product data that matches the search keyword entered by the user 3 is less than the specified number, it is determined that the product data search has failed, and the reference keyword is extracted. You may make it do.

  At this time, the product search unit 101 of the product search device 1 determines that the search for the product data has failed when the search result of the product data that matches the search keyword input by the user 3 is equal to or less than the predetermined number, The reference keyword extraction unit 102 is requested to extract the reference keyword, the product data and the reference keyword included in the search result are delivered to the web server unit 100, and the product data and the reference keyword included in the search result are displayed. Request generation.

  In this way, by setting the input error judgment criterion to be equal to or less than the predetermined number instead of 0, even if product data corresponding to the search keyword with the input error is searched, product related terms that may be correct Can be presented to the user.

  Of course, the products handled in the present invention include personal computers, computer peripherals, home appliances, sports equipment, books, music software, video software, Internet providers, real estate, vehicles (cars, motorcycles, bicycles, etc.), It may be a food item, or a service such as finance, insurance, or travel reservation.

  As described above, according to the product search device and the product search method according to the present embodiment, when the search for product data using the search keyword input by the user to the terminal device fails, the similarity with the search keyword is determined. Product-related terms that meet the specified criteria are extracted as reference keywords, and the extracted reference keywords are displayed on the terminal device, so that the product-related terms that are likely to be correct for the input keyword are presented to the user. It is possible to compensate for input errors due to lack of knowledge and carelessness of the user.

  For example, if the product search device according to the present embodiment is applied to a shopping site, the user misunderstands that there is no product on the shopping site when the search result by the search keyword input by the user to the shopping site is zero. , Shopping site separation (the user ending use of the shopping site) can be prevented.

The figure explaining embodiment of a goods search device. The functional block diagram of a goods search apparatus and a client. The figure explaining the web page containing the form which inputs a search keyword. The figure explaining the specific example of a 1st calculation method. The figure explaining the web page on which a reference keyword is displayed. The flowchart which showed the procedure of the goods search method.

Explanation of symbols

1 Product Search Device 100 Web Server Unit 101 Product Search Unit 102 Reference Keyword Extraction Unit 110 Product Database 111 Product Related Word Database

Claims (11)

  A product search method for searching for product data that matches a search keyword input by a user, wherein the device includes a product database in which product data including product-related words that are words related to the product is stored. When the product data search request is received from the terminal device, the device searches the product data including the search keyword transmitted from the terminal device as the product related word from the product database, a, When the device determines that the product data search has failed in step a, the device obtains the similarity between the product-related word stored in the product database and the search keyword from the search keyword. The number of the first partial character string to be included in the product related word is divided by the total number of the first partial character strings. And the second partial character string obtained from the product related word is calculated by multiplying the number obtained by dividing the number of the second partial character string contained in the search keyword by the total number of the second partial character strings, and the similarity is determined. A step b for extracting the product related words that meet the determined criteria as a reference keyword, the device generates display data for displaying the reference keyword extracted in the step b on the terminal device, and the terminal device A merchandise search method comprising: transmitting step c.   The product search method according to claim 1, wherein the product-related word is generated using a word included in a specification or a description of a product stored in the product database. Product search method.   The product search method according to claim 1 or 2, wherein in the step b, the apparatus uses partial character strings having different numbers of characters as the partial character strings.   The product search method according to any one of claims 1 to 3, wherein in the step b, the device is configured such that the number of the product data including the search keyword as the product related word is 0. Alternatively, when the number is equal to or less than a predetermined number, the device determines that the device has failed in the search for the product data.   The product search method according to any one of claims 1 to 4, wherein the device includes the product database for each category of the product, and the step a is a terminal device operated by the user. When the product data search request is received from the device, the device includes the product data including the search keyword transmitted from the terminal device as the product-related word in the product category transmitted from the terminal device. In the step of searching from the corresponding product database, when the device determines that the search of the product data has failed in the step a, the device is configured to search for the product transmitted from the terminal device. Calculating the similarity between the product-related terms stored in the product database corresponding to the category and the search keyword; Product search wherein the similarity score is a step of extracting reference keyword matching the items related words to criteria have been established.   A product search device for searching for product data that matches a search keyword input by a user, the product database storing product data including product related words that are words related to the product, and a terminal device operated by the user When a product data search request is received, product search means for searching the product database for the product data including the search keyword transmitted from the terminal device as the product related word; and the product search means is the product data The number of first partial character strings obtained from the search keyword indicating the degree of similarity between the product-related word stored in the product database and the search keyword when the search fails. Divided by the total number of the first partial character strings and a second partial character string obtained from the product-related words Multiplying the number contained in the search keyword by the value obtained by dividing the number by the total number of the second partial character strings, and extracting the product-related words that meet the criteria for determining the similarity as reference keywords Reference keyword extraction means, and reference keyword display means for generating display data for displaying the reference keyword extracted by the reference keyword extraction means on the terminal device, and transmitting the generated display data to the terminal device. A product search device characterized by that.   The product search device according to claim 6, wherein the product-related word is generated using a word included in a specification or an explanatory text of a product stored in the product database. Product search device. A product search device according to claim 6 or claim 7, wherein the reference keyword extracting means, as the partial string, product search apparatus characterized by utilizing a partial character string of the different characters.   The product search device according to any one of claims 6 to 8, wherein the product search means includes 0 items of the product data including the search keyword as the product related word, or A product search device that determines that the search for the product data has failed when the number is equal to or less than a predetermined number.   The product search device according to any one of claims 6 to 9, wherein the product search device is a web server arranged on the Internet, and the reference keyword display unit includes the product search unit. A product search device for generating HTML data for displaying the reference keyword in the form of a hyperlink in which a URL of the product search device to be operated is embedded. The product search device according to any one of claims 6 to 10, wherein the product search device includes the product database for each category of the product, and the product search unit is operated by the user. When the product data search request is received from the terminal device, the product data including the search keyword transmitted from the terminal device as the product related word corresponds to the category of the product transmitted from the terminal device. The reference keyword extracting means is a means for searching from the product database, wherein the reference keyword extracting means stores the product database corresponding to the product category transmitted from the terminal device when the product search means fails to search the product data. Calculation of similarity between stored product related words and search keyword, and determination of similarity Product search device, characterized in that the means for extracting quasi in conforming the items related word by reference keyword.

Images