Search systems, search methods, and programs

Abstract

In the search system, the search results for products similar to the search criteria are presented in a way that makes it easy for the user to compare and consider them. [Solution] The search system comprises a product information database that stores product information for each product, a vector database that stores vector information obtained by vectorizing the product information for each product, a reception unit that receives input of search conditions from the user, a search unit that searches for products similar to the input search conditions by vector search by referring to the vector database, an extraction unit that extracts items that differ from the input search conditions for each of the searched products by referring to the product information database, and an output processing unit that represents the extracted differing items in characters and outputs the search results from the search unit.

Description

Technical Field

[0001] The present invention relates to a search system, a search method, and a program.

Background Art

[0002] In a product search site or the like for searching for products, in addition to products that match the search conditions specified by the user, a similarity search may be used to search for products that do not match but are similar. For example, in Patent Document 1, a desired level of each attribute of a product is input, a similarity between each input desired level and each attribute level of each product is obtained, and using the similarity for each obtained attribute, a fit rate of each product with respect to a desired product having the input desired level is calculated, and based on the calculated fit rate, a recommendation system that displays one or more similar products as recommended products is disclosed.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the technique described in Patent Document 1 above, in a product database, the attribute levels of each product are stored numerically, and it is difficult to perform a similarity search unless the product specifications or the like can be compared numerically. In response to such a problem, it is conceivable to use vector search that quantifies the meaning of data and searches for similar information. However, in vector search, it is difficult for the user to grasp the basis of similarity such as which feature amount (for example, specifications, etc.) of the product is similar to the search conditions. Therefore, in order for the user to grasp what is different between the searched similar products and the search conditions, it is necessary to compare and check the product specifications or the like.

[0005] This invention has been made in view of these circumstances, and its purpose is to provide a search system, a search method, and a program that can present the results of searching for products similar to the search criteria in a manner that is easy for the user to compare and consider. [Means for solving the problem]

[0006] To solve the above-mentioned problems, one aspect of the present invention provides a product information database that stores product information for each product, and product information for each product The text data contained within is input into a trained vector generation model to create an embedding vector. The system comprises a vector database that stores vector-converted vector information, a reception unit that receives input of search conditions from the user, a search unit that searches for products similar to the input search conditions by vector search by referring to the vector database, an extraction unit that extracts items that differ from the input search conditions for each of the searched products by referring to the product information database, and an output processing unit that represents the extracted differences in characters and outputs the search results from the search unit. The extraction unit refers to the product information database and extracts the values ​​of pre-set important items from the items included in the product information for each of the searched products, and the output processing unit represents the extracted values ​​of important items in the search results. It is a search system.

[0007] Furthermore, one aspect of the present invention is a search method performed by a computer used in a search system, wherein the computer receives input of search conditions from a user and retrieves product information for each product. The text data contained within is input into a trained vector generation model to create an embedding vector. The system searches for products similar to the input search criteria by referring to a vector database that stores vector-converted vector information, and then searches for products similar to the input search criteria by referring to a product information database that stores product information for each product. For each of the searched products, it extracts the differences from the input search criteria, represents the extracted differences in text, and outputs the search results. A search method that includes extracting items that differ from the input search conditions includes referring to the product information database and extracting the values ​​of pre-set important items from the items included in the product information for each of the searched products, and outputting the search results includes representing the values ​​of the extracted important items in the search results. This is the search method.

[0008] Furthermore, one aspect of the present invention involves a computer used in a search system that receives input of search conditions from a user. hair, Product information for each item The text data contained within is input into a trained vector generation model to create an embedding vector.Referencing a vector database that stores vector-converted vector information, search for products similar to the input search criteria using vector search. death, Referencing a product information database that stores product information for each product, extract any discrepancies between the search criteria entered and the searched product for each of the searched products. death, The extracted differences are represented in text, and the search results are output. The program performs the following actions: Extracting items that differ from the input search conditions includes referring to the product information database and, for each of the searched products, extracting the values ​​of pre-set important items from the items included in the product information; and outputting the search results includes representing the values ​​of the extracted important items in the search results. It is a program. [Effects of the Invention]

[0009] As described above, according to this invention, the search system can present the search results for products similar to the search criteria in a manner that makes it easy for the user to compare and consider them. [Brief explanation of the drawing]

[0010] [Figure 1] This is a schematic block diagram showing the system configuration of a search system according to an embodiment. [Figure 2] This is a schematic block diagram showing a specific example of the functional configuration of a terminal device according to the embodiment. [Figure 3] This is a schematic block diagram showing a specific example of the functional configuration of a search device according to an embodiment. [Figure 4] This figure shows a specific example of product information stored in the product information database according to the embodiment. [Figure 5] This figure shows a specific example of vector information stored in a vector database according to the embodiment. [Figure 6] This flowchart shows a specific example of the search process performed by the search device according to the embodiment. [Figure 7] This figure shows an example of a product details screen according to the embodiment. [Figure 8] This figure shows a first example of a search results screen according to the embodiment. [Figure 9] This figure shows a second example of the search results screen according to the embodiment. [Figure 10]It is a flowchart showing a specific example of a search process for recursively searching for similar products executed by the search device according to the embodiment. [Figure 11] It is a diagram showing a modified example of the search device.

Mode for Carrying Out the Invention

[0011] Hereinafter, a search system, a search method, and a program according to an embodiment of the present invention will be described with reference to the drawings.

[0012] <Configuration of Search System 100> FIG. 1 is a schematic block diagram showing the system configuration of the search system 100 according to the present embodiment. The search system 100 is an information processing system that provides a product search site for searching for a product required by a user from a group of products (for example, products, job offers, housing, etc.). The search system 100 includes a terminal device 10 and a search device 20. The terminal device 10 and the search device 20 are communicably connected via a network N. The network N may be a network using wireless communication or a network using wired communication. The network N may be any one of the Internet, a public wireless communication network (PLMN: Public Land Mobile Network), a local area network (LAN), a virtual private network (VPN), or any combination thereof.

[0013] <Configuration of Terminal Device 10> FIG. 2 is a schematic block diagram showing a specific example of the functional configuration of the terminal device 10 according to the present embodiment. The terminal device 10 is an information terminal operated by a user. The terminal device 10 is configured using an information device such as a smartphone, a tablet, a personal computer, or a dedicated device. The terminal device 10 includes a communication unit 11, an input unit 12, a display unit 13, a storage unit 14, and a control unit 15.

[0014] The communication unit 11 is a communication device. The communication unit 11 may be configured, for example, as a network interface. The communication unit 11 communicates data with other devices via the network N in accordance with the control of the control unit 15. The communication unit 11 may be a device that performs wireless communication or a device that performs wired communication.

[0015] The input unit 12 is configured using existing input devices such as a keyboard, pointing device (mouse, tablet, etc.), buttons, or touch panel. The input unit 12 is operated by the user when inputting user instructions to the terminal device 10. The input unit 12 may also be an interface for connecting the input device to the terminal device 10. In this case, the input unit 12 inputs the input signal generated in the input device in response to the user's input to the terminal device 10. The input unit 12 may also be configured using a microphone and a speech recognition device. In this case, the input unit 12 acquires the acoustic signal generated by the user's speech, performs speech recognition on the words spoken by the user, and inputs the recognized string information to the terminal device 10. The speech recognition process may be performed by the control unit 15. The input unit 12 may be configured in any way that allows user instructions to be input to the terminal device 10.

[0016] The display unit 13 displays information in a format that is recognizable to the user. The display unit 13 may be an image display device such as a liquid crystal display or an organic EL (Electro-Luminescence) display. The display unit 13 may also be an interface for connecting the image display device to the terminal device 10. In this case, the display unit 13 generates a video signal for displaying image data and outputs the video signal to the image display device connected to it. The display unit 13 may also be configured as a touch panel integrated with the input unit 12.

[0017] The storage unit 14 has the function of storing various types of information. The storage unit 14 is composed of storage media provided as hardware by the terminal device 10, such as an HDD (Hard Disk Drive), SSD (Solid State Drive), flash memory, EEPROM (Electrically Erasable Programmable Read Only Memory), RAM (Random Access Memory), ROM (Read Only Memory), or any combination of these storage media. The storage unit 14 stores data used by the control unit 15. The storage unit 14 stores data necessary when the control unit 15 performs processing.

[0018] The control unit 15 is composed of a processor such as a CPU (Central Processing Unit) and memory (main memory). The control unit 15 functions when the processor executes a program. Note that all or part of the functions of the control unit 15 may be implemented using hardware such as an ASIC (Application Specific Integrated Circuit), PLD (Programmable Logic Device), or FPGA (Field Programmable Gate Array). The above program may be recorded on a computer-readable recording medium. Computer-readable recording media include, for example, portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and semiconductor memory devices (e.g., SSDs: Solid State Drives), as well as storage devices such as hard disks and semiconductor memory devices built into computer systems. The above program may be transmitted via a telecommunications line.

[0019] The control unit 15 may, for example, execute an application installed on its own device (terminal device 10). A specific example of such an application is an application provided to the terminal device 10 as a dedicated application for the search system 100. Another specific example of such an application is a web browser application. Such an application may be pre-installed on the terminal device 10, or it may be downloaded each time a search process, as described later, is performed. For example, if it is implemented as a web browser application, the terminal device 10 may download and execute the application from a device specified by the web server (for example, the web server itself or another server) when the terminal device 10 connects to a specific web server. The control unit 15 operates according to the program of the application being executed.

[0020] The control unit 15 controls the terminal device 10 in response to user operations and information received from the search device 20. For example, the control unit 15 transmits information entered by the user through the input unit 12 to the search device 20 using the communication unit 11. For example, when the control unit 15 receives information transmitted from the search device 20 via the network N to the communication unit 11, it generates screen data based on the received information and displays the screen data on the display unit 13. Such screen data includes images and characters that represent the information transmitted from the search device 20.

[0021] <Configuration of the search device 20> Figure 3 is a schematic block diagram showing a specific example of the functional configuration of the search device 20 according to this embodiment. The search device 20 is a device that searches for products desired by the user. The search device 20 is configured using, for example, one or more information processing devices such as personal computers or server devices. The search device 20 may also be a cloud server located on a so-called cloud. The search device 20 comprises a communication unit 21, a storage unit 22, and a control unit 23.

[0022] The communication unit 21 is a communication device. The communication unit 21 may be configured, for example, as a network interface. The communication unit 21 communicates data with other devices via the network N in accordance with the control of the control unit 23. The communication unit 21 may be a wireless communication device or a wired communication device.

[0023] The storage unit 22 has the function of storing various types of information. The storage unit 22 is composed of storage media provided as hardware by the search device 20, such as an HDD, SSD, flash memory, EEPROM, RAM, ROM, or any combination of these storage media. The storage unit 22 stores data used by the control unit 23. The storage unit 22 may function, for example, as a product information database 221, a vector database 222, and a vector generation model storage unit 223.

[0024] The product information database 221 stores product information for each of the multiple products being searched. The product information includes at least the product's specifications or features.

[0025] (Product Information Database) Figure 4 shows a specific example of product information stored in the product information database 221 according to this embodiment. The product information database 221 has multiple product information records. Each product information record contains product information and includes values ​​such as product ID, model number, type, product name, and specifications (in the illustrated example, body, bonnet / cap / cover, etc.). The product ID is identification information that uniquely identifies the product. The model number is the model number of the product. The type is the type of product. The product name is the product name of the product. The specifications are the specifications or features of the product. Product information may also include information such as the product category, an image representing the appearance of the product, and the price of the product.

[0026] Returning to Figure 3, the vector database 222 stores vector information obtained by vectorizing each of the product information items for multiple products.

[0027] (Vector database) Figure 5 shows a specific example of vector information stored in the vector database 222 according to this embodiment. The vector database 222 has multiple vector information records. Each vector information record represents vector information and includes, for example, a product ID and the values ​​of the vectors. The product ID is identification information that uniquely identifies a product. The vectors are vectorized product information. The number of vectors stored in each vector information record may be one or two or more (for example, vector V1, vector V2, ...). For example, the database may be configured to contain two or more vectors for each product, such as "spec vector" and "related document vector (for example, FAQ document)".

[0028] Returning to Figure 3, the vector generation model storage unit 223 stores the vector generation model that the vector transformation unit 232 (described later) uses to transform data into vectors. The vector generation model is an AI model or rule-based transformation model that takes text or image data as input and outputs a vector obtained by transforming the input data into vectors. Examples of the former include Sentence-BERT and OpenAI Embeddings, while examples of the latter include the Bag of Words method and the TF-IDF method, but are not limited to these. Furthermore, the vector generation model may be, for example, an embedding model. The vector generation model may be constructed using information from a pre-trained model generated by a training process based on product information stored in the product information database 221. Such a training process may be performed by another device, for example, or by the device itself (search device 20). For example, the vector generation model may be fine-tuned to perform vector transformations suitable for the categories of the product group (the product group to be searched) stored in the product information database 221.

[0029] The control unit 23 is configured using a processor such as a CPU and memory. The control unit 23 functions as a receiving unit 231, a vector conversion unit 232, a search unit 233, an extraction unit 234, an output processing unit 235, and an update unit 236, as the processor executes a program. Note that all or part of the functions of the control unit 23 may be implemented using hardware such as an ASIC, PLD, or FPGA. The above program may be recorded on a computer-readable recording medium. Computer-readable recording media include, for example, portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and semiconductor storage devices (e.g., SSDs), as well as storage devices such as hard disks and semiconductor storage devices built into computer systems. The above program may also be transmitted via a telecommunications line.

[0030] The reception unit 231 receives search criteria for products from the user. For example, the reception unit 231 may display an input screen for entering search criteria on the display unit 13 of the terminal device 10. Alternatively, the reception unit 231 may receive a specification of a specific product from the user and use the product information of the specified product (hereinafter sometimes referred to as "specified product") as the search criteria.

[0031] The vector transformation unit 232 transforms the input search conditions into vectors. For example, the vector transformation unit 232 inputs the search conditions into a vector generation model and obtains the vectorized search conditions output from the vector generation model.

[0032] The search unit 233 searches the vector database 222 for product IDs of similar products that are similar to the vectorized search conditions. At this time, the search unit 233 obtains the similarity between the search conditions and similar products by calculating the similarity between the vectors. For example, cosine similarity or the Jacquard coefficient method may be used to calculate the similarity between the vectors. Furthermore, if the vector consists of multiple or higher-dimensional quantities, weighting of each vector or element may be considered.

[0033] The extraction unit 234 reads the product information for each similar product found from the product information database 221, and, referring to the retrieved product information, extracts differences between the search conditions and each similar product (e.g., specifications, features, etc.). The extraction unit 234 also extracts values ​​for pre-set important items from the retrieved product information. Important items are, for example, items that users consider important when comparing products. Important items may be pre-set for each product category or type.

[0034] The output processing unit 235 outputs the search results from the search unit 233, representing the differences between the search criteria and similar products, as well as the values ​​of important items, in text. For example, the output processing unit 235 displays the search results screen on the display unit 13 of the terminal device 10. Alternatively, the output processing unit 235 may transmit the search results data to the terminal device 10 or other server devices via the communication unit 11.

[0035] The update unit 236 updates the vector information stored in the vector database 222 based on the updated product information when there is an update to the product information in the product information database 221. In addition, if there is an update in the product information database 221 that exceeds a predetermined amount, the update unit 236 may perform additional training of the vector generation model based on the updated product information (for example, added product information).

[0036] <Processing flow of the search device 20> Next, with reference to Figures 6 to 9, the processing flow of the search device 20 according to this embodiment will be described. Figure 6 is a flowchart showing a specific example of the search process performed by the search device 20 according to this embodiment. Below, the case of searching for products will be described using the example of a valve as the search target. In this example, the vector generation model is fine-tuned to pre-learn product information for the product category "valve" and output a vector based on the learned product information.

[0037] First, the reception unit 231 receives input of search conditions from the user (step S101). For example, the reception unit 231 may display an input screen on the display unit 13 of the terminal device 10 for entering search conditions (for example, filtering conditions specifying parameters such as product specifications and features). Alternatively, the reception unit 231 may receive instruction input from the user to search for similar products to a specified product on a product search site, either on a product details screen that displays individual product information (for example, detailed specifications and features) or on a product list screen that displays a list of products.

[0038] (Product details screen) Figure 7 shows an example of a product details screen 700 according to this embodiment. This figure shows the product details screen 700 for a product with model number "(A)XXXXXX". As shown in the figure, the product details screen 700 includes a similarity search button 710 that accepts an instruction input to search for similar products of that product. When the reception unit 231 receives input from the user to press the similarity search button 710 on the product details screen 700, it designates that product (model number "(A)XXXXXX") as the specified product. The reception unit 231 then retrieves the product information of the specified product (model number "(A)XXXXXX") from the product information database 221 and uses the retrieved product information (for example, specifications, etc.) as search conditions. Note that which of the values ​​of each item included in the product information will be used as search conditions may be set in advance.

[0039] Returning to Figure 6, the vector transformation unit 232 then transforms the search conditions into vectors by inputting them into the vector generation model (step S102). Next, the search unit 233 uses the vectorized search conditions to perform a vector search on the vector database 222 for product IDs of similar products that are similar to the search conditions (specified product) (step S103). At this time, the search unit 233 calculates the similarity between the search conditions (specified product) and the similar products (for example, the cosine similarity between the vectors). The search unit 233 may consider products with a similarity score above a preset threshold as similar products.

[0040] Next, the extraction unit 234 retrieves product information corresponding to each product ID of the searched similar product from the product information database 221 (step S104). Next, the extraction unit 234 compares each acquired product information with the search conditions (product information of the specified product) to extract items that differ from the search conditions (specified product) for each similar product (for example, items representing specifications or details) (step S105). Next, the extraction unit 234 extracts the values ​​of important items for each similar product from the acquired product information (step S106). For example, the extraction unit 234 extracts the valve type and the body type as important items.

[0041] Next, the output processing unit 235 outputs the search results (step S107). For example, the output processing unit 235 displays the search results screen on the display unit 13 of the terminal device 10. After that, the search device 20 terminates the search process.

[0042] (Example 1 of the search results screen) Figure 8 shows a first example of the search results screen 800 according to this embodiment. The search results screen 800 includes a table 810 that lists similar products to the search criteria (specified product). The output processing unit 235 displays similar products with a higher similarity score calculated in the vector search in the table 810, prioritizing them (for example, at the top or highlighted). In the illustrated example, the output processing unit 235 sorts the similar products in the table 810 in descending order of similarity (so that similar products with a higher similarity score are displayed at the top). For example, the output processing unit 235 displays the specified product (model number "(A)XXXXXX") on the first row of the table 810, and displays the similar product with the highest similarity to the specified product (model number "(A)XXXAB") (similarity 98.7%) on the second row.

[0043] Table 810 contains the following items: similarity, image, model number / link, main differences, valve, and body. The output processing unit 235 displays the similarity calculated in the vector search in the similarity column of Table 810. The output processing unit 235 also displays the image of each similar product in the image column of Table 810. Furthermore, the output processing unit 235 displays the model number of each similar product in the model number / link column of Table 810. The displayed model number may include a link to the product details screen of the similar product. This allows the user to access the product details screen of the corresponding product by clicking or touching the model number.

[0044] Furthermore, the output processing unit 235 displays the differences between the search criteria (specified product) and similar products in text in the "Main Differences" column of the table 810. For example, in the illustrated example, the output processing unit 235 displays the difference "internal thread" as the main difference between the specified product (model number "(A)XXXXXX") and the similar product (model number "(B)YYYYYY"). In this way, by displaying the main differences between the specified product and each similar product in text on the search results screen 800, the user can check the differences between the specified product and each similar product at a glance. In other words, the user can grasp the differences (e.g., differences in specifications) between the specified product and each similar product listed on the search results screen 800 without having to access the product details screen to check the specifications and other details of each similar product, and can easily compare and consider the products. Therefore, the search device 20 can help improve the efficiency of product searches by the user. The output processing unit 235 may display all differences in the product information as the main differences, or it may prioritize displaying pre-set priority items as the main differences. For example, the output processing unit 235 may display only the values ​​of priority items among the extracted differences as the main differences.

[0045] Furthermore, the output processing unit 235 displays the valve type of each similar product in the valve column of table 810. The output processing unit 235 also displays the body type of each similar product in the body column of table 810. Valves and bodies are pre-set important items. By displaying the values ​​of important items (valves and bodies) that users consider important when comparing products, it is possible to further improve the efficiency of product searches. In the illustrated example, the important items are fixed to valves and bodies, but this is not limited to this; users may specify important items on the search results screen 800. For example, the output processing unit 235 may allow users to select important items using a pull-down menu or similar on the search results screen 800. In this case, the output processing unit 235 retrieves the values ​​of the important items selected by the user from the product information database 221 and displays the retrieved values ​​on the search results screen 800. This allows users to arbitrarily select the items they consider important when comparing products and to view their values ​​all at once on the search results screen 800.

[0046] Furthermore, the output processing unit 235 may sort the similar products displayed in table 810 in ascending or descending order based on the values ​​of important items. For example, the output processing unit 235 may sort the similar products and update the display in table 810 in response to user input (for example, input such as pressing the valve item name 811 or the body item name 812) so that similar products with the same valve type or body type as the specified product (matching the search criteria) are displayed at the top. In the illustrated example, for example, if the user inputs input to press the valve item name 811, the output processing unit 235 may sort the similar products and update the display in table 810 so that similar products with the valve type "gate valve" (same as the specified product) are displayed at the top. The output processing unit 235 may also group and display the search results based on the values ​​of important items.

[0047] Furthermore, the output processing unit 235 may display the product information (i.e., search criteria) 820 of the specified product on the search results screen 800. This allows the user to easily understand the search criteria (product information of the specified product) for similar products.

[0048] Furthermore, the output processing unit 235 may accept instructions from the user to perform a further similar search from the search results screen 800.

[0049] (Second example of the search results screen) Figure 9 shows a second example of the search results screen 800 according to this embodiment. As shown in the figure, the output processing unit 235 may display a similarity search button 830 on the search results screen 800 corresponding to each similar product in the search results. The similarity search button 830 receives an instruction input from the user to search for similar products of the corresponding product. When the search device 20 receives an operation input from the user to press the similarity search button 830, it performs a vector search for similar products of the corresponding product and updates the display on the search results screen 800 based on the search results obtained from the vector search. In other words, the similarity search button 830 can be said to be an example of an implementation form corresponding to the search condition input reception (step S101) in Figure 6. That is, the search device 20 may provide a user interface that allows recursive similarity searches to be called from the search results. The flowchart in that case is shown in Figure 10. The processing from steps S201 to S207 shown in Figure 10 is the same as the processing from steps S101 to S107 shown in Figure 6, so its explanation is omitted. When the similarity search button 830 is pressed on the search results screen 800, the control unit 23 sets the product information of the corresponding product as a search condition (step S208) and returns to the process in step S202. In other words, by recursively displaying products similar to those for which the search conditions have been relaxed or which did not match the original conditions, it also provides an opportunity to discover products that were not found.

[0050] As described above, the search system 100 according to this embodiment includes a product information database 221 that stores product information for each product, a vector database 222 that stores vector information obtained by vectorizing the product information for each product, a reception unit 231 that receives input of search conditions from the user, a search unit 233 that performs a vector search for products similar to the input search conditions by referring to the vector database 222, an extraction unit 234 that extracts items that differ from the input search conditions for each of the searched products by referring to the product information database 221, and an output processing unit 235 that represents the extracted differing items in characters and outputs the search results from the search unit 233.

[0051] With this configuration, the search system 100 according to this embodiment uses vector search to search for products similar to the search conditions, making it possible to perform similarity searches even if the specifications to be compared are not necessarily numerical. On the other hand, the similarity between vectors obtained through vector search is an abstract numerical value, making it difficult for humans to understand the basis for it. To address this problem, the search system 100 according to this embodiment outputs the search results by expressing the differences between the search conditions and the found similar products in text, thereby clearly indicating the differences between the search conditions and similar products and providing the user with easy-to-understand "vector search results." As a result, the user can refer to the search results and check what the differences are between the search conditions and each similar product at once, making it easier to compare and consider products. Therefore, the search system 100 can present the search results for products similar to the search conditions in a manner that is easy for the user to compare and consider. In this way, the search system 100 can help improve the efficiency of product searches by the user.

[0052] Furthermore, the search system 100 according to this embodiment refers to the product information database 221 and, for each product searched, extracts the values ​​of pre-set important items from the items included in the product information, and displays the extracted values ​​of important items in the search results. In this way, by displaying the values ​​of important items that users consider important when comparing products in the search results, the search system 100 can further help improve the efficiency of product searches.

[0053] Furthermore, the search system 100 according to this embodiment accepts product specifications from the user and uses the product information of the specified product as search criteria. With this configuration, the search system 100 according to this embodiment can search for products similar to any product specified by the user. Therefore, it eliminates the need for the user to input search criteria (for example, detailed product specifications, etc.) and makes similarity searches easier. In other words, the specified product can be rephrased as a product that 100% matches the search criteria, meaning that all parameters must match those criteria. To put this in general terms, even if a product does not 100% match the "refining criteria," if this UI is equipped to display the differences that are different from the matching criteria, then where conventional search results would show "no results found," only the conditions that do not match can be clearly indicated and displayed.

[0054] Furthermore, the search system 100 according to this embodiment displays the extracted differences in text in the search results and shows a list of the searched products. With this configuration, the search system 100 according to this embodiment can present the differences between the search conditions and each searched product to the user in a list. As a result, the user can check all at once what each product differs from the search conditions, further improving search efficiency.

[0055] Furthermore, the search system 100 according to this embodiment prioritizes displaying products with high similarity in vector search results. With this configuration, the search system 100 according to this embodiment can prioritize presenting products that are more similar to the search criteria to the user (for example, through recommendations), making it easier for the user to find the product they are looking for.

[0056] In the embodiment described above, the search device 20 stores vector information of product information in the vector database 222. However, it is not limited to this, and information such as FAQs (Frequently Asked Questions) on the search site may also be stored as vector information in the vector database 222 in addition to product information. This allows, for example, when a user enters search conditions in natural language, the device to present the user with products that closely match the search conditions. Furthermore, although the above-described embodiment was explained using the example of searching for products on a product search site, the present invention is not limited to this and can also be applied to comparison search sites for job postings (e.g., employment or career change), housing, etc.

[0057] (modified version) In this embodiment, the terminal device 10 and the search device 20 are configured as separate devices, but they may be configured as a single integrated device. Figure 11 shows a modified example of the search device 20 configured in this way. The search device 20 shown in Figure 11 includes an input unit 24 and a display unit 25. The input unit 24 and display unit 25 of the search device 20 shown in Figure 11 function similarly to the input unit 12 and display unit 13 of the terminal device 10, respectively. The control unit 23 operates in response to operations on the input unit 24, performs search processing using the input information, and outputs the information using the display unit 25.

[0058] Furthermore, the search device 20 may be implemented using multiple information processing devices. For example, the search device 20 may be implemented using a cloud or other similar device. For example, in the search device 20, the storage unit 22 and the control unit 23 may be implemented in different information processing devices. For example, the product information database 221, vector database 222, and vector generation model storage unit 223 of the storage unit 22 of the search device 20 may be distributed and implemented across multiple information processing devices.

[0059] Some or all of the functions of the search device 20 in the above-described embodiment may be implemented by a computer. In that case, the functions may be implemented by recording a program for implementing these functions on a computer-readable recording medium, loading the program recorded on this recording medium into a computer system, and executing it. Here, "computer system" includes hardware such as an OS and peripheral devices. Furthermore, "computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and storage devices such as hard disks built into a computer system. Moreover, "computer-readable recording medium" may also include those that dynamically hold programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or communication lines such as telephone lines, and those that hold programs for a certain period of time, such as volatile memory inside a computer system that acts as a server or client in such a case. Furthermore, the above-mentioned program may be for implementing some of the functions described above, or it may be a program that can implement the above-mentioned functions in combination with a program already recorded in the computer system, or it may be implemented using a programmable logic device such as an FPGA (Field Programmable Gate Array).

[0060] Although embodiments of this invention have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments and includes designs and the like that do not depart from the spirit of this invention. [Explanation of symbols]

[0061] 100 Search Systems 10 Terminal devices 11 Communications Department 12 Input section 13 Display section 14 Storage section 15 Control Unit 20 Search device 21 Communications Department 22 Memory section 221 Product Information Database 222 Vector Databases 223 Vector generation model memory 23 Control Unit 231 Reception Department 232 Vector Transformation Unit 233 Search Section 234 Extraction part 235 Output Processing Unit 236 Update Department

Claims

1. A product information database that stores product information for each product, A vector database that stores vector information obtained by inputting text data contained in the product information of each product into a trained vector generation model and converting it into embedded vectors, A reception area that receives search criteria input from users, A search unit that searches for products similar to the input search criteria by referring to the aforementioned vector database, An extraction unit that, by referring to the aforementioned product information database, extracts items that differ from the input search conditions for each of the searched products, An output processing unit that represents the extracted differences in text and outputs the search results from the search unit, Equipped with, The extraction unit refers to the product information database and, for each of the searched products, extracts the values ​​of the pre-set important items from the items included in the product information. The output processing unit, in the search results, represents the values ​​of the extracted important items. Search system.

2. The reception unit receives product specifications from the user and uses the product information of the specified product as the search criteria. The search system according to claim 1.

3. The output processing unit displays the extracted differences in text in the search results and lists the searched products. The search system according to claim 1.

4. The output processing unit prioritizes displaying products with high similarity in the vector search in the search results. The search system according to claim 1.

5. The output processing unit displays an operator that accepts an operation from the user to search for products similar to the products included in the search results, corresponding to each of the multiple products included in the search results. When the search unit receives an operation from the user for the operator, it searches for products similar to the product corresponding to the operator for which the operation was received, by referring to the vector database and performing a vector search. The search system according to claim 1.

6. A search method performed by a computer used in a search system, The system accepts search criteria input from the user. The system searches for products similar to the input search criteria by referring to a vector database that stores vector information obtained by inputting the text data contained in the product information of each product into a trained vector generation model and converting it into embedded vectors. Referencing a product information database that stores product information for each product, extract any discrepancies between the search criteria entered and the searched product for each of the searched products. The extracted differences are represented in text, and the search results are output. This is a search method that includes the following: Extracting items that differ from the input search conditions includes referring to the product information database and, for each of the searched products, extracting the values ​​of pre-set important items from the items included in the product information. Outputting the search results includes representing the values ​​of the extracted important items in the search results. How to search.

7. The computers used in the search system The system accepts search criteria input from the user. The system searches for products similar to the input search criteria by referring to a vector database that stores vector information obtained by inputting the text data contained in the product information of each product into a trained vector generation model and converting it into embedded vectors. Referencing a product information database that stores product information for each product, extract any discrepancies between the search criteria entered and the searched product for each of the searched products. The extracted differences are represented in text, and the search results are output. It is a program that makes something happen. Extracting items that differ from the input search conditions includes referring to the product information database and, for each of the searched products, extracting the values ​​of pre-set important items from the items included in the product information. Outputting the search results includes representing the values ​​of the extracted important items in the search results. program.

Images