JP7571464B2 - Search server, search method and program - Google Patents

Description

The present invention relates to technology for searching for information on the Internet.

As smartphones have become more widespread, SNS (Social Network Systems) have become established in society, with many people posting photos of food and information about the restaurants they have eaten at. Young people who were born and raised in an environment where smartphones are widespread, the so-called smartphone native generation, are increasingly using SNS when searching for restaurants, rather than traditional gourmet sites.

Patent document 1 describes a system that allows users to easily search for useful stores and other information using the Internet.

JP 2019-215843 A

However, the method of Patent Document 1 does not take trends or popularity into consideration at all in the search. Currently, trends in food, fashion, and the like change quickly, and information on popular restaurants tends to change in a short period of time. As a result, searches using conventional gourmet sites can result in outdated information that is not useful. Even if you use SNS, there is an abundance of information on SNS, making it difficult to find the restaurant you want to go to from the many restaurants available. Furthermore, even if you see an image of food posted on SNS and think, "I want to go," many posts do not include information about the restaurant, such as the name or address of the restaurant.

The present invention has been made to solve the problems described above, for example, and its main objective is to provide a search server that searches the Internet in response to a specific input and provides users with recommended store information.

One aspect of the present invention is a search server communicatively connected to a network and a user terminal, the search server including a receiving unit that receives text and images relating to a product from the user terminal as input information, a popular post extraction unit that extracts posts similar to the input information from the network, calculates the popularity of the extracted posts, and extracts posts with high popularity as popular posts, a similarity calculation unit that calculates a similarity between an image of a product included in the input information and an image included in the popular post, and if there is a popular post including an image whose similarity is equal to or greater than a threshold, determines the image of the product included in the input information to be a specific image, and if there is no popular post including an image whose similarity is equal to or greater than the threshold, determines the image of the product included in the input information to be a generic image, and identifies a category of the product based on the generic image. The system includes a category identification unit, a store post extraction unit that extracts from the popular posts, based on the input information, posts that include product information about the product and store information that is information for purchasing the product, as store posts, a store list creation unit that creates a store list displaying information about one or more stores based on the store posts, and a transmission unit that transmits the store list to the user terminal, and if there is a popular post including an image whose similarity is greater than or equal to a threshold, the store post extraction unit extracts from the popular posts, as the store post, posts that include an image whose similarity to the specific image is greater than or equal to a threshold and the store information, and if there is no popular post including an image whose similarity is greater than or equal to the threshold, extracts from the popular posts, as the store post, posts that include product information representing the category of the product and the store information .

The above search server checks how similar the product image included in the popular post is to the product image input by the user, and determines whether the product image input by the user is a specific image or a generic image based on the similarity. If there is a popular post including an image with a similarity equal to or greater than a threshold, the search server extracts a popular post including an image with a similarity equal to or greater than a threshold with respect to the specific image and store information as a store post. According to this aspect, it is possible to extract store posts and create a store list without identifying a product category. On the other hand, if there is no popular post including an image with a similarity equal to or greater than a threshold, the search server identifies a product category based on a generic image, and extracts a post including product information representing the product category and store information from the popular post as a store post. In this aspect, it is possible to determine whether the product image input by the user is a generic image representing a product in a certain category, and to identify the product category based on the generic image. In this way, the search server extracts posts including product information and store information from the popular posts based on input information received from the user terminal, and transmits the information to the user terminal as a store list. This allows users to obtain and view a list of stores that have been carefully selected from popular posts, simply by making simple input about the products they are interested in on their user terminal.

In one aspect of the above search server, the receiving unit receives location information of the user terminal, and the store list creation unit identifies the address of the store based on the store information, and calculates one or more of the route, travel time, or fare from the store's address and the location information to obtain information about the store. In this aspect, the user can easily check the route, travel time, fare, etc. to a store by viewing the store list.

In another aspect of the search server, the store list creation unit calculates a route that a user will take to visit multiple stores based on the store posts, and sets the calculated route as information about the stores. In this aspect, a user can easily check a route to visit multiple stores by viewing the store list, which is useful, for example, when visiting multiple stores to collect goods.

In another aspect of the search server, the popular post extraction unit calculates the popularity based on one or more of the following indicators: the number of views, the number of citations, the number of uses, the number of reposts, and the number of likes for the post, and a post with a popularity equal to or greater than a threshold is designated as a popular post. In this aspect, the popularity is calculated based on one or more of the following indicators: the number of views, the number of citations, the number of uses, the number of reposts, and the number of likes for the post, so that a so-called buzzworthy post that has suddenly become a hot topic and attracted the attention of many people on the Internet, particularly SNS, can be designated as a popular post.

In another aspect of the search server, the store list creation unit creates the store list based on store posts that include images whose similarity to the specific image is equal to or greater than a threshold. In this aspect, if the image input by the user is a specific image, the user may be interested in "that product at that store," so the store list is created to include store information corresponding to images similar to the specific image. This makes it possible to provide a high-quality store list that meets the user's wishes.

In another aspect of the search server, the store list creation unit rearranges the display order of multiple stores based on one or more of the following indicators: the distance from the user's current location to the store, the required time or fare, the price of the product, the popularity of the store post, and the similarity of the image. In this aspect, the display order of the store list can be easily rearranged based on the indicator desired by the user.

The popular post extraction unit extracts popular posts within a specified period of time. In this embodiment, the popular posts to be extracted can be determined according to the period of time, so that, for example, posts that have been a hot topic in the past can be excluded and only posts that have become a hot topic in a short period of time can be determined as popular posts.

In another aspect of the present invention, there is provided a search method executed by a search server communicatively connected to a network and a user terminal, the search method comprising the steps of: receiving text and images relating to a product from a user terminal as input information; extracting posts similar to the input information from the network; calculating a popularity of the extracted posts; extracting posts with high popularity as popular posts; calculating a similarity between an image of a product included in the input information and an image included in the popular post; determining the image of the product included in the input information to be a specific image if there is a popular post including an image whose similarity is equal to or greater than a threshold; determining the image of the product included in the input information to be a generic image if there is no popular post including an image whose similarity is equal to or greater than a threshold; and selecting a category of the product based on the generic image. The search server identifies a target store and performs a store post extraction process to extract, from the popular posts, posts including product information about the product and store information for purchasing the product as store posts based on the input information, creates a store list displaying information about one or more stores based on the store posts, and transmits the store list to the user terminal, and in the case where there is a popular post including an image whose similarity is equal to or greater than a threshold, the store post extraction process extracts, from the popular posts, posts including an image whose similarity to the specific image is equal to or greater than a threshold and the store information as the store posts, and in the case where there is no popular post including an image whose similarity is equal to or greater than a threshold, extracts, from the popular posts, posts including product information representing the product category and the store information as store posts . By executing this search method, the search server can transmit information about stores carefully selected from the popular posts as a store list to the user terminal.

In yet another aspect of the present invention, there is provided a program executed by a computer communicatively connected to a network and a user terminal, the program including a receiving unit that receives text and images relating to a product from the user terminal as input information, a popular post extraction unit that extracts posts similar to the input information from the network, calculates a popularity of the extracted posts, and extracts posts with high popularity as popular posts, a similarity calculation unit that calculates a similarity between an image of a product included in the input information and an image included in the popular post, and if there is a popular post including an image whose similarity is equal to or greater than a threshold, determines the image of the product included in the input information to be a specific image, and if there is no popular post including an image whose similarity is equal to or greater than the threshold, determines the image of the product included in the input information to be a generic image, and a similarity calculation unit that identifies a category of the product based on the generic image. The program causes the computer to function as a category specifying unit that specifies a category of the product to be included in the popular posts, a store post extraction unit that extracts, from the popular posts, posts including product information about the product and store information that is information for purchasing the product based on the input information as store posts, a store list creation unit that creates a store list displaying information about one or more stores based on the store posts, and a transmission unit that transmits the store list to the user terminal, and the store post extraction unit extracts, if there is a popular post including an image whose similarity is equal to or greater than a threshold, posts including an image whose similarity to the specific image is equal to or greater than a threshold and the store information as the store posts from the popular posts, and if there is no popular post including an image whose similarity is equal to or greater than the threshold, posts including product information representing the product category and the store information as store posts from the popular posts . The above-mentioned search server can be realized by executing this program on a computer.

The search server according to the present invention can search the Internet in response to a specific input and provide the user with recommended store information.

1 shows a configuration of a search system according to an embodiment. FIG. 2 is a block diagram showing an internal configuration of a search server. 13 is an example of a store list display. Examples of information to include in your store listing 4 is a flowchart of a search process in the first embodiment. 13 is a flowchart of a search process in the second embodiment. 13 is a display example of a store list in the first modified example.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
First Embodiment
[Overall configuration]
1 shows the configuration of a search system to which the search server of the present invention is applied. The search system 100 is configured so that a search server 1 capable of searching the Internet for a specific input and extracting necessary information, and a user terminal 3 which is a terminal device used by a user, can communicate with each other via a network 5. Here, the network 5 includes the Internet. The Internet is a huge communication network constructed by interconnecting multiple networks around the world for the purpose of transmitting information.

For the sake of convenience, FIG. 1 shows only one user terminal 3 used by one user, but in an actual search system 100, the search server 1 is configured to be able to communicate with multiple user terminals used by multiple users. Furthermore, communication between the search server 1 and the user mobile phone 3 via the network 5 may be wireless or wired, as shown in FIG. 1.

The search server 1 can communicate with the user terminal 3 via the network 5, and receives text and images related to the product from the user terminal 3 as input information. The search server 1 then uses a web (World Wide Web) service provided on the Internet to search for posts that include store information for the product. Hereinafter, searches and extractions using web services provided on the Internet will be referred to as "searching the Internet" and "extracting from the Internet." In addition, SNSs, blogs (Weblogs), and various websites on web services provided on the Internet will be referred to as "SNSs, blogs, and various websites on the Internet."

Here, store information is information for purchasing a product, such as the name of the store that provides the product (hereinafter also referred to as the "store name"), the address or location information, the brand name of the product, etc., and can be set arbitrarily. Stores include companies and websites that sell products on the Internet. Furthermore, posts are text and images published on the Internet, such as SNS including Twitter (registered trademark), Instagram, Facebook, Line, etc., blogs, various websites, etc. The search server 1 creates a store list including the store information of the product based on the search results, and transmits it to the user terminal 3.

The user terminal 3 is a variety of terminal devices used by users, such as smartphones, tablets, desktop PCs, and laptop PCs, and transmits text and images related to products input by the user as input information to the search server 1. The user terminal 3 also receives and displays a store list from the search server 1. By viewing the store list, the user can obtain store information for the product. Based on the store information, the user can, for example, go to eat the product if it is food, or purchase the product if it is clothing or goods. In this embodiment, the user terminal 3 is a smartphone.

[Search Server Configuration]
2 is a block diagram showing the internal configuration of the search server 1. The search server 1 includes a communication unit 21, a popular post extraction unit 22, a category identification unit 23, a store post extraction unit 24, a similarity calculation unit 25, and a store list creation unit 26. These components, a user information DB 28, and classifier information 29 are connected to each other via a bus 30.

The communication unit 21 is a communication unit for communicating with the user terminal 3 via the network 5. Specifically, the communication unit 21 receives text and images related to products from the user terminal 3 as input information, and transmits the created store list to the user terminal 3. If necessary, the communication unit 21 also receives location information of the user terminal 3 together with the input information or separately.

The user information DB 28 stores user information about users. The user information is, for example, an ID and an address required to send the created store list to the user terminal 3. Specifically, the user registers in advance using the user terminal 3, and the search server 1 stores the information acquired during the user registration as user information for each user.

In this embodiment, the user is required to register in advance, but the present invention is not limited to this, and as long as the store list created by the search server 1 can be appropriately transmitted to the user terminal 3, user registration and the user information DB 28 are not necessary.

The classifier information 29 stores information about a classifier for identifying product images included in the input information. Here, the classifier is a learning model that is trained to output a classification result related to the category of a product when an image of the product is input. This learning model may be any learning model used in machine learning, such as a support vector machine or a neural network. The classifier is trained using training data that is, for example, a combination of product images of various categories and ground truth data indicating the correct category of the product. This training data includes a sufficient number of the above combinations for each category that the classifier should classify. The training of the classifier may be performed by the search server 1, or may be executed by a device other than the search server 1.

Specifically, the classifier identifies the category of a product from the product image included in the input information, and assigns a label indicating the identified category to each product image. In other words, the search server 1 inputs the product image included in the input information to a classifier configured based on the classifier information 29, thereby obtaining a label indicating the category of the product image. The search server 1 then identifies the product category from the obtained label.

In this embodiment, the search server 1 has the classifier information 29, but the present invention is not limited to this. As long as the category of a product can be identified from an image of the product, another server that can communicate with the search server 1 may have the classifier information 29, or an image recognition service may be used. In other words, any method can be applied to identify the category of a product from an image of the product included in the input information.

The popular post extraction unit 22 extracts posts that are creating a buzz on the Internet, primarily on SNS, as popular posts. To go viral means that a topic spreads all at once and attracts the attention of many people. Specifically, the popular post extraction unit 22 calculates the buzz level (hereinafter also referred to as "popularity") based on indicators such as the number of views (impressions), number of views, number of citations, number of uses, number of reposts, and number of likes of a post, and extracts posts with a popularity level equal to or greater than a threshold value from the Internet as popular posts. The popularity level may be calculated based on one indicator, or may be calculated by a composite data analysis based on multiple indicators.

In addition, since there are many posts on the Internet, the popular post extraction unit 22 may extract posts similar to the input information from the Internet, calculate the popularity of the extracted posts, and extract posts whose popularity is equal to or exceeds a threshold value as popular posts.

The category identification unit 23 identifies the category of the product based on the text and image related to the product included in the input information. Here, the text related to the product is a character string representing the category of the product handled in the store, such as "shaved ice" for food and "one piece" for clothing. Also, the image related to the product is an image showing a product such as "shaved ice" or "one piece". When the input information is text representing a product, the category identification unit 23 recognizes the text and identifies the category of the product. On the other hand, when the input information is an image of a product, the category identification unit 23 identifies the category of the product using a classifier configured based on the classifier information 29 as described above.

The store post extraction unit 24 extracts, from the popular posts extracted by the popular post extraction unit 22, posts that include product information representing the product input by the user as input information and store information for purchasing the product as store posts. Specifically, the product information is information representing the category of the product identified by the category identification unit 23. For example, if the product category is "shaved ice," the store post extraction unit 24 checks whether the text and images included in the popular post contain information representing "shaved ice." Furthermore, the store post extraction unit 24 checks whether the popular post that contains information representing "shaved ice" contains store information such as the name of the store where the "shaved ice" was purchased. In this way, the store post extraction unit 24 extracts, from the popular posts, posts that contain both product information and store information as store posts.

The similarity calculation unit 25 calculates the similarity between the product image included in the input information and the image included in the post. One example of a similarity calculation method is to extract features from the image using a CNN (Convolution Neural Network) and calculate the similarity using cosine similarity. Note that the present invention is not limited to this, and a feature extractor other than CNN may be used, and a similarity calculation function other than cosine similarity may be used. Furthermore, other similarity calculation methods such as template matching and pattern matching may be used. In other words, any method can be applied as the similarity calculation method by the similarity calculation unit 25.

The similarity calculation unit 25 also calculates the similarity between the image of the product included in the input information and the image included in the store post, and determines whether the image of the product is a generic image or a specific image based on the similarity. Here, a generic image is an image that represents a product in a certain category, and a specific image is an image that shows a specific product in that category that is handled by a specific store. For example, if the product is "shaved ice," the generic image is an image of general shaved ice regardless of the store, and the specific image is an image of a specific shaved ice offered at a certain popular store. Specifically, when there is a store post including an image with a similarity equal to or greater than a threshold, the similarity calculation unit 25 determines the image of the product to be a specific image, and when there is no store post including an image with a similarity equal to or greater than a threshold, the similarity calculation unit 25 determines the image of the product to be a generic image.

The store list creation unit 26 creates a store list that displays information about one or more stores based on the store posts extracted by the store post extraction unit 24. Here, the store-related information includes store information for purchasing products as well as additional information such as routes to the stores, which will be described later.

Specifically, the store list creation unit 26 first identifies store information for purchasing products, such as the store name, address, official website URL (Uniform Resource Locator), and product price, based on the extracted store posts. Here, an official website is a website that a company or store has set up to introduce its products or brands. At this time, the store list creation unit 26 identifies the store name, address, official website URL, and the like from the text and images included in the store posts, but if, for example, only the store name can be identified and the address and official website URL are unknown, the store list creation unit 26 may search the Internet from the store name to identify the address and official website URL.

The store list creation unit 26 also calculates the distance, route, required time, fare, etc. from the current location to the store based on the identified address and the location information of the user terminal 3. The store list creation unit 26 then edits store information such as the store name, address, official website URL, and product price of each store identified based on the store posts, as well as additional information such as the calculated distance, route, required time, and fare from the current location to each store, to create a store list as information about each store.

The store list creation unit 26 may calculate the route, required time, fare, etc. from the current location to multiple stores, instead of the route from the current location to each store, based on the extracted store posts. In this case, the store list creation unit 26 includes the calculated route, required time, fare, etc. from the current location to multiple stores in the store list and provides it to the user.

In addition, when a user inputs an image of a product, the store list creation unit 26 creates a store list that includes store information corresponding to images similar to the specific image if the image of the product is a specific image, since there is a possibility that the user "wants to eat that product at that store." Specifically, the store list creation unit 26 creates a store list based on store posts that include images whose similarity to the specific image is equal to or greater than a threshold. At this time, the store list creation unit 26 may include images whose similarity to the specific image is equal to or greater than a threshold in the information about the corresponding store.

The display order of each store in the store list is determined based on indicators such as the distance from the user's current location to the store, the travel time or fare, the price of the product, the popularity of the store posts corresponding to the store, and the similarity of the images corresponding to the store. In other words, the display order of stores in the store list may be rearranged based on any indicator, such as "shortest travel time," "most popular," or "most similar image." Furthermore, the store list creation unit 26 may create the store list after ranking stores based on any indicator. The display order and ranking may be determined based on one indicator, or may be determined by a composite data analysis based on multiple indicators.

Here, the store list will be described with reference to Figures 3 and 4. Figure 3 is an example of a store list. Figure 4 is an example of information to be included in the store list.

The store list has a rank 50, a store name 51, an address 52, a website 53, a route 54, and a required time 55. In the example of FIG. 3, the store list is created for three stores, Store A, Store B, and Store C, with store names 51, and each store's address 52, website 53, route 54, and required time 55 as store-related information. The website 53 is linked to the URL of each store's official website, and is created so that the official website of each store is displayed when pressed on the user terminal 3. The route 54 is created so that the distance, route, and fare from the user's current location to the store are displayed when pressed on the user terminal 3. The rank 50 is the order in which the stores are arranged in order of "shortest required time" using the required time 55 of each store as an index, and the store list displays information about each store in order of rank from top to bottom.

The store list may also display information showing a route from the current location to multiple stores, as shown in FIG. 4(a). The store list may also display information about the corresponding store, including images whose similarity to a specific image is equal to or greater than a threshold, as shown in FIG. 4(b).

In this way, the store list creation unit 26 searches the Internet based on store posts, identifies store information and calculates various information related to multiple stores, and then creates a store list in which the information is organized by store for easy viewing by users. Note that the store-related information that makes up the store list can be set arbitrarily, but the text and images used must be within the scope of copyright and other regulations and restrictions.

In the above configuration, the communication unit 21 of the search server 1 is an example of the receiving unit and transmitting unit of the present invention. In addition, the popular post extraction unit 22, category identification unit 23, store post extraction unit 24, similarity calculation unit 25, and store list creation unit 26 of the search server 1 are examples of the popular post extraction unit, category identification unit, store post extraction unit, similarity calculation unit, and store list creation unit of the present invention, respectively.

[Search process]
5 is a flowchart of the search process in the first embodiment. This process is a process of extracting store posts by identifying a product category based on input information and creating a store list. Specifically, this process is realized by the computer constituting the search server 1 executing a program prepared in advance.

First, the user transmits text and images related to a product that the user is interested in or is willing to purchase to the search server 1 as input information using the user terminal 3. The input information may be either text or images related to the product, or both.

When the search server 1 receives input information from the user terminal 3 (step S401), it determines whether the information contained in the input information is only an image of a product (step S402). If the information contained in the input information is only an image of a product (step S402; Yes), the search server 1 identifies the category of the product based on the image of the product (step S403). On the other hand, if the information contained in the input information is not only an image of the product (step S402; No), the search server 1 identifies the category of the product by recognizing the text (step S404).

Next, the search server 1 extracts posts that are creating a buzz on the Internet, mainly on SNS, as popular posts (step S405). Specifically, the search server 1 calculates popularity based on a predetermined index such as the number of views, and extracts posts from the Internet whose popularity is equal to or exceeds a threshold as popular posts. The search server 1 then extracts, from the popular posts, posts that contain both product information indicating the product category and store information for purchasing the product as store posts (step S406). Furthermore, the search server 1 determines whether the input information includes an image of the product (step S407). If there is no image of the product (step S407; No), the search server 1 proceeds to the process of step S412.

On the other hand, if there is an image of the product (step S407; Yes), the search server 1 calculates the similarity between the image of the product and the image included in the store post (step S408). Furthermore, the search server 1 determines whether there is a store post including an image whose calculated similarity is equal to or greater than a threshold (step S409). If there is a store post including an image whose similarity is equal to or greater than a threshold (step S409; Yes), the search server 1 determines the product image included in the input information to be a specific image (step S410). On the other hand, if there is no store post including an image whose similarity is equal to or greater than the threshold (step S409; No), the search server 1 determines the product image included in the input information to be a generic image (step S411).

Then, the search server 1 creates a store list based on the store post (step S412). Specifically, the search server 1 searches the Internet based on the store post, identifies store information and calculates various information related to multiple stores, and then creates a store list in which information is organized by store for easy viewing by the user. At this time, when the user inputs an image of a product, the search server 1 creates a store list that includes store information corresponding to images similar to the specific image, since there is a possibility that the user "wants to eat that product from that store" contained in the image, if the image of the product is a specific image. The search server 1 then transmits the created store list to the user terminal 3 (step S413). This ends the search process.

Second Embodiment
[Search process]
Next, the search process in the second embodiment will be described. In the second embodiment, the input information includes an image of a product. The overall configuration and the configuration of the search server are similar to those in the first embodiment, and therefore the description will be omitted. FIG. 6 is a flowchart of the search process in the second embodiment. This process is a process of extracting store posts that include an image similar to an image of a product, and creating a store list. Specifically, this process is realized by the computer constituting the search server 1 executing a program prepared in advance.

First, the search server 1 receives input information including an image of a product from the user terminal 3 (step S501). The search server 1 also extracts posts that are creating a buzz on the Internet, mainly on SNS, as popular posts (step S502). The search server 1 then calculates the similarity between the product image included in the input information and the image included in the popular post (step S503). Furthermore, the search server 1 determines whether there is a popular post that includes an image with a calculated similarity equal to or greater than a threshold (step S504).

If there is a popular post containing an image with a similarity equal to or greater than the threshold (step S504; Yes), the search server 1 determines that the image of the product contained in the input information is the specific image (step S505). Furthermore, the search server 1 extracts, from the popular posts, posts containing both an image with a similarity equal to or greater than the threshold with the specific image and store information for purchasing the product as store posts (step S506). Next, the search server 1 determines whether there is a corresponding store post (step S507), and if there is a corresponding store post (step S507; Yes), creates a store list based on the store post (step S511). At this time, when the user inputs an image of a product, the search server 1 creates a store list so as to include store information corresponding to an image similar to the specific image, since there is a possibility that the user "wants to eat that product at that store" contained in the image. Note that in the processing of step S507, if there is no corresponding store post (step S507; No), the search server 1 proceeds to processing of step S509.

On the other hand, if there is no popular post containing an image with a similarity equal to or greater than the threshold (step S504; No), the search server 1 determines that the image of the product contained in the input information is a generic image (step S508). Then, the search server 1 identifies the category of the product based on the image of the product contained in the input information (step S509). Furthermore, the search server 1 extracts, from the popular posts, posts that contain both product information indicating the product category and store information for purchasing the product as store posts (step S510). Then, the search server 1 creates a store list based on the store posts (step S511).

Once the store list has been created, the search server 1 transmits the store list to the user terminal (step S512). This ends the search process. In this way, in the search process of the second embodiment, the search server 1 can extract store posts based on the product image included in the input information, without first identifying the product category. Then, if there is no image in the popular post that has a similarity to the product image equal to or greater than a threshold value, or if there is an image in the popular post but the popular post does not contain store information, the search server 1 extracts store posts by identifying the product category from the product image.

In this way, according to the search system 100 of the present invention, a user can obtain and view information about stores carefully selected from popular posts as a store list by simply inputting information about a product of interest on the user terminal 3. In other words, a user can obtain information about stores that are the center of topics and trends or famous stores that are always attracting attention with simple operations, without having to search the Internet, which is overflowing with information, by themselves.

Furthermore, if a user wants to purchase a product they saw on television or social media but does not know the necessary purchasing information, such as the store, brand name, or product name, they can easily obtain store information, such as the store name, address, and official website, needed to purchase the product. Furthermore, users can also obtain additional information, such as the price of the product, the time it takes to get to the store, and routes to visit multiple stores. Therefore, users can effectively use the store list even when they are not familiar with the area, for example, when they are traveling.

<Modification>
(First Modification)
In the first and second embodiments, the popular post extraction unit 22 of the search server 1 may set a period and extract popular posts within the period. The period may be set arbitrarily, for example, within one week. In this way, posts that have recently gone viral can be extracted as popular posts, rather than posts that have gone viral in the past.

The popular post extraction unit 22 may also calculate the popularity by analyzing the time series of a period. For example, if a given post previously had a low number of views, but the number of views has increased dramatically within the past week, a possible reason for this may be that the post was "broadcast on television." Therefore, by calculating the popularity by analyzing the time series of a period, the popular post extraction unit 22 can extract posts that are currently at the center of ongoing topics or trends as popular posts.

Furthermore, when a post whose popularity has risen sharply in the most recent short period of time compared to past popularity is extracted as a store post, a buzzing icon 56 indicating that the store's popularity is rising sharply may be displayed for the store corresponding to the store post in the store list, as shown in FIG. 7.

This allows users to view information about stores that are currently the center of attention or trends, and the icon 56 allows them to distinguish these stores from stores that have been the center of attention in the past or are standard stores.

(Second Modification)
In the case of text, the input information about a product may include not only the product category but also other information. For example, if the product is clothing, it may be input as "a dress worn by celebrity XX" or "a jacket seen in drama XX." In this case, the search server 1 treats the celebrity's name "XX" or the drama title "XX" as product information together with the product category "dress," and creates a store list by extracting store posts from popular posts.

(Third Modification)
When an image of a product is used as input information, the user may specify whether the image of the product is a specific image or a generic image on the user terminal 3. Specifically, a check box may be provided on the screen when submitting input information, and the user may check the box to specify whether the image is a specific image or a generic image. This allows the search server 1 to recognize whether the user means "I want to eat that product at that store" contained in the product image, or whether the product image is just listed as an example of that category.

Furthermore, if the image of a product is a specific image but there is no store post containing that image and the store list creation unit 26 cannot provide store information for purchasing the product, the store list creation unit 26 may also display a message such as "The product with the image you are looking for could not be found" in the store list.

(Fourth Modification)
When the store list creation unit 26 identifies location information from a store post as store information, the location information may be dummy, so the store list creation unit 26 may set a judgment condition for using the location information in the store list. For example, the judgment condition may be that more than half of the location information extracted from multiple store posts is the same.

(Fifth Modification)
In the search system 100, the network 5 includes the Internet, but the present invention is not limited to this, and as long as the search server 1 is connected to the Internet, it is sufficient that communication is possible between the search server 1 and the user terminal 3. In this case, the network 5 does not have to include the Internet.

Reference Signs List 1 Search server 3 User terminal 5 Network 22 Popular post extraction unit 23 Category identification unit 24 Store post extraction unit 25 Similarity calculation unit 26 Store list creation unit 100 Search system

Claims (9)

A search server communicably connected to a network and a user terminal,
A receiving unit that receives text and images related to the product from a user terminal as input information;
a popular post extraction unit that extracts posts similar to the input information from the network, calculates a popularity of the extracted posts, and extracts posts with high popularity as popular posts;
a similarity calculation unit that calculates a similarity between an image of a product included in the input information and an image included in the popular post, and determines the image of the product included in the input information to be a specific image when there is a popular post including an image whose similarity is equal to or greater than a threshold, and determines the image of the product included in the input information to be a generic image when there is no popular post including an image whose similarity is equal to or greater than a threshold;
a category identification unit that identifies a category of the product based on the general-purpose image;
a store post extraction unit that extracts, from the popular posts, posts including product information related to the product and store information for purchasing the product based on the input information, as store posts;
A store list creation unit that creates a store list displaying information about one or more stores based on the store posts;
A transmission unit that transmits the store list to the user terminal;
Equipped with
The store post extraction unit is a search server that, if there is a popular post including an image whose similarity to the specific image is above a threshold, extracts from the popular posts posts that include an image whose similarity to the specific image is above a threshold and the store information as the store post, and if there is no popular post including an image whose similarity to the specific image is above a threshold, extracts from the popular posts posts that include product information representing the product category and the store information as the store post . The receiving unit receives location information of the user terminal,
The search server of claim 1, wherein the store list creation unit identifies the address of the store based on the store information, and calculates one or more of the route, travel time, or fare from the store's address and the location information to obtain information about the store. The search server according to claim 2, wherein the store list creation unit calculates a route that a user will take to visit multiple stores based on the store posts, and generates information about the stores. The search server according to any one of claims 1 to 3, wherein the popular post extraction unit calculates the popularity based on one or more of the following indicators: the number of views, the number of citations, the number of uses, the number of reposts, and the number of likes of the post, and defines posts whose popularity is equal to or exceeds a threshold as popular posts. The search server according to any one of claims 1 to 4, wherein the store list creation unit creates the store list based on store posts that include images whose similarity to the specific image is equal to or greater than a threshold value. The search server according to any one of claims 1 to 5, wherein the store list creation unit rearranges the display order of multiple stores based on one or more indicators of the distance from the user's current location to the store, the required time or fare , the price of the product, the popularity of the store post, and the similarity of the image. The search server according to claim 1 , wherein the popular post extraction unit extracts popular posts within a predetermined period. A search method executed by a search server communicably connected to a network and a user terminal,
Receive text and images related to the product as input information from the user terminal;
extracting posts similar to the input information from the network, calculating the popularity of the extracted posts, and extracting posts with high popularity as popular posts;
Calculate a similarity between an image of a product included in the input information and an image included in the popular post, and if there is a popular post including an image with the similarity equal to or greater than a threshold, determine the image of the product included in the input information to be a specific image, and if there is no popular post including an image with the similarity equal to or greater than a threshold, determine the image of the product included in the input information to be a generic image;
Identifying a category of the product based on the generic image;
performing a store post extraction process for extracting, from the popular posts, posts including product information related to the product and store information that is information for purchasing the product, as store posts based on the input information;
Based on the store post, a store list is created that displays information about one or more stores;
Transmitting the store list to the user terminal;
The store post extraction process is a search method in which, if there is a popular post including an image whose similarity is above a threshold, a post that includes an image whose similarity to the specific image is above a threshold and the store information is extracted from the popular post as the store post, and if there is no popular post including an image whose similarity is above a threshold, a post that includes product information representing the product category and the store information is extracted as the store post from the popular post . A program executed by a computer communicatively connected to a network and a user terminal,
A receiving unit that receives text and images related to the product from a user terminal as input information;
a popular post extraction unit that extracts posts similar to the input information from the network, calculates a popularity of the extracted posts, and extracts posts with high popularity as popular posts;
a similarity calculation unit that calculates a similarity between an image of a product included in the input information and an image included in the popular post, and determines the image of the product included in the input information to be a specific image when there is a popular post including an image whose similarity is equal to or greater than a threshold, and determines the image of the product included in the input information to be a generic image when there is no popular post including an image whose similarity is equal to or greater than a threshold;
a category identification unit that identifies a category of the product based on the general-purpose image;
a store post extraction unit that extracts, from the popular posts, posts including product information related to the product and store information for purchasing the product based on the input information, as store posts;
A store list creation unit that creates a store list displaying information about one or more stores based on the store posts;
a transmission unit that transmits the store list to the user terminal ;
The store post extraction unit is a program that, when there is a popular post including an image whose similarity is above a threshold, extracts from the popular posts posts that include an image whose similarity to the specific image is above a threshold and the store information as the store post, and when there is no popular post including an image whose similarity is above a threshold, extracts from the popular posts posts that include product information representing the product category and the store information as the store post .