A product identification and classification method based on human-computer interaction and chat robot

Abstract

The application relates to a product identification classification method based on human-computer interaction and a chat robot, and the method comprises the following steps: step 1: loading a configuration file of a database configured with product classification relations; step 2: loading a first question and a plurality of answer options corresponding to the question and sending the question and the answer options to a user, and receiving the answer returned by the user; step 3: continuously sending a next question and corresponding answer options to the user until there is no next question corresponding to the answer returned by the current user, and matching all the answers returned by the user with product categories in the database according to product classification relations. The application does not need to listen to and negotiate through a server, and can meet the needs of more application scenarios. The application only needs to modify corresponding data without changing codes, occupies less storage space, is easy to maintain and modify, and does not need additional learning costs.

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction chatbot technology, specifically a product identification and classification method and chatbot based on human-computer interaction. Background Technology

[0002] In certain scenarios, such as e-commerce shopping platforms, users often remember a product but are unclear about its category, making it difficult to quickly find the purchase link and leading to failed purchases. In such cases, users want the platform to guide them to the product's category, enabling them to view and purchase it. To achieve this, e-commerce platforms need to interact with users to determine the true category of the product they wish to buy. This can be done through chatbots, which can interact with users to identify and categorize products. Chatbots can be physical hardware terminals or virtual software.

[0003] Existing chatbots offer product identification and categorization functions, typically integrating this functionality into a website. Users access a specific URL and answer questions to categorize products. However, this solution is limited by the inability to easily integrate the functionality into other front-end platforms, resulting in a poor user experience and limited scalability. Furthermore, the Q&A interaction often involves users initiating questions, with the website (system) providing the answers. This process, coupled with users' limited product knowledge, frequently leads to low efficiency and even failed product categorization. Summary of the Invention

[0004] To address the shortcomings of existing technologies, one of the objectives of this invention is to provide a product identification and classification method based on human-computer interaction, which can solve the problem of efficiently identifying product categories during human-computer interaction.

[0005] A second objective of this invention is to provide a chatbot that can solve the problem of efficiently identifying product categories during human-computer interaction.

[0006] One of the technical solutions to achieve the objective of this invention is: a product identification and classification method based on human-computer interaction, comprising the following steps:

[0007] Step 1: Load the configuration file, which contains a database of product category relationships;

[0008] Step 2: Load the first question and its corresponding answer options from the database in the configuration file and send them to the user, then receive the answer options returned by the user;

[0009] Step 3: Continue sending the user the next question and its corresponding answer options until the current user's answer does not have a corresponding next question.

[0010] Based on all the answers returned by the user, the system matches them with product categories in the database according to product classification relationships. If a product category is found, the user is informed that a suitable product category has been found; otherwise, the user is informed that a suitable product category cannot be found.

[0011] Furthermore, in step 3, after receiving a unique answer from the user for the current question, the answer is matched with the answer options for the same question in the database. If a match is found, it is checked whether there is a next question under that answer option. If so, the next question is sent to the user, and all the answers returned by the user are matched with the product categories in the database according to the product classification relationship.

[0012] If not, then directly match all the existing returned answers with the product categories in the database based on the product classification relationships.

[0013] Furthermore, in the database of the configuration file, a table is used to store the product classification relationship logic. Each table cell in the top row of the table stores a question, and each table cell in the first column starting from the second row stores a product category. Each table cell to the right of each product category stores an answer corresponding to each question. Answers corresponding to the same question in the same column constitute the answer options.

[0014] Furthermore, the first question in the table is loaded, and the user is sent with the question and the answer options corresponding to the column containing the question. An answer is then returned from the answer options.

[0015] After receiving the answer returned by the user, move the next table cell to the right from the row containing the answer. If the next table cell is not empty, move the column upwards to find the first question in the column containing the current answer, and send the found question and its corresponding answer options to the user again. Continue until the table cell to the right of the found answer is empty or does not exist. Then, send all questions to the user and obtain all the answers returned by the user.

[0016] Furthermore, after step 1 and before step 2, the following steps are also included:

[0017] Obtain user data, predict the shortest classification path problem in the database based on the user data, and take the shortest classification path problem as the first problem.

[0018] The second technical solution to achieve the objective of this invention is: a chatbot, comprising,

[0019] Memory, used to store program instructions;

[0020] A processor is configured to run the program instructions to perform the steps of the product identification and classification method based on human-computer interaction.

[0021] The beneficial effects of this invention are as follows: By dynamically loading a question-answer database, modifying the answer to a question or changing the dialogue logic only requires modifying the corresponding data in the database, without changing the code. Furthermore, using tables to store data occupies less storage space compared to storing data in a tree structure, and is easier to maintain and modify without requiring additional learning costs. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the process of the present invention;

[0023] Figure 2 This is a schematic diagram of a chatbot. Detailed Implementation Plan

[0024] The present invention will now be further described with reference to the accompanying drawings and specific embodiments:

[0025] like Figure 1 As shown, a product identification and classification method based on human-computer interaction includes the following steps:

[0026] Step 1: Load the configuration file, which contains a database of product classification relationships. In actual use, when human-computer interaction is needed for product identification and classification, the product classification function object is first awakened by the scheduler. Then, the configuration file is loaded onto the product classification function object, storing information such as database interfaces. Based on the information in the configuration file, the product classification relationships are loaded into the database in memory. The database stores all the question-and-answer logic, and subsequent dialogue will be based entirely on the content in the database.

[0027] Before this step and step 2, i.e., before loading the first question, user data is first obtained. This user data typically includes the user's location (usually at the city / region level), browsing history, consumption history, purchase history, product category usage history, etc., and is input into the scheduler. Then, based on this user data, it predicts which question leads to the shortest classification path. The question leading to the shortest classification path is selected as the first question, i.e., the first question mentioned in step 2. The scheduler can be configured with models including decision trees, support vector machines, neural networks, etc., forming a comprehensive learning model. The user data is then input into this comprehensive learning model for prediction to determine which question leads to the shortest classification path.

[0028] Step 2: Load the first question and its corresponding answer options from the database in the configuration file, send the first question and its corresponding answer options to the user, and receive the user's unique answer option. For each product identification and classification function question, there are multiple corresponding answer options. The question and answer options are pre-stored in the database. Multiple answer options are sent to the user, who selects a specific answer and returns it.

[0029] In this step, after multiple dialogues and responses with the user, the length of the question-answer path for each dialogue, from the first question to the last specific answer, can be collected. This path length is then fed back into the user data, forming a feedback mechanism. This mechanism allows for continuous updating and improvement of the user data, leading to more accurate predictions of the first question. In the short term, the feedback mechanism shortens the classification path for a single user; in the long term (e.g., once a week), this user data can be used to retrain the machine learning model in the scheduler, helping to shorten the product classification question path length (for all users).

[0030] Step 3: Check if there is a next question. If so, continue sending questions and corresponding answer options to the user until the current answer returned by the user does not have a corresponding next question. Based on all the answers returned by the user, match them with the product categories in the database according to the product classification relationship. If a product category information can be matched, that is, the product category can be determined based on the answer returned by the user, inform the user that a suitable product category has been found and provide the product category. Otherwise, inform the user that a suitable product category cannot be found.

[0031] In this step, after receiving a unique answer from the user for the current question, the answer is matched against the answer options for the same question in the database. If a match is found, it checks if there is a next question under that answer option. If so, it continues to send the next question to the user and matches all the answers returned by the user against the product categories in the database based on product classification relationships.

[0032] If not, then directly match all the existing returned answers with the product categories in the database based on the product classification relationships.

[0033] Referring to Table 1, the product category relationship logic is stored in a table within the configuration file's database. The top row of the table stores several questions, with each cell allowing only one question. The first column of the second row stores product categories, with each cell storing only one product category. To the right of each product category, the table cells store the answer options for each question. Each cell stores one answer, and multiple answers combine to form the answer options for a question; that is, answers for the same question in the same column constitute the answer options. This setup allows each question to have multiple answer options, which users can choose to answer. When a user answers multiple questions and selects the corresponding answer, it uniquely corresponds to the leftmost product category.

[0034] Question 1 Question 2 Question 3 Product Category 1 Answer 10 Answer 20 Answer 30 Product Category 2 Answer 11 Answer 21 Answer 31 Product Category 3 Answer 12 Answer 22 Answer 32 Product Category 4 Answer 13 Answer 23 Answer 33 Question 4 Question 5 Product Category 5 Answer 14 Answer 24 Answer 34 Answer 40 Answer 50 Product Category 6 Answer 15 Answer 25 Answer 35 Answer 41 Answer 51 Product Category 7 Answer 16 Answer 26 Answer 36 Answer 42 Answer 52 …… Question n Product Category 8 Answer 17 Answer 27 Answer 37 Answer 43 Answer 53 Answer n0 Product Category 9 Answer 18 Answer 28 Answer 38 Answer 44 Answer 54 Answer n1 Product Category 10 Answer 19 Answer 29 Answer 39 Answer 45 Answer 55 Answer n2

[0035] Table 1

[0036] From Table 1, Question 1 has 10 answers, from answer 10 to answer 19. These 10 answers form the answer options for Question 1. Similarly, Question 2 also has 10 answers, which form the answer options for Question 2.

[0037] The system first sends question 1 and its corresponding answer options to the user. The user selects a specific answer and returns it, for example, answer 11. After receiving the user's returned answer, the system moves the next table cell to the right from the row containing that answer. If the next table cell is not empty, the system moves upward from the column containing the current answer to search for the first question (which is now question 2) and sends the found question to the user again. Then, the user continues to select an answer from the answer options in the column containing question 2, and the search continues until the table cell to the right of the found answer is empty or does not exist. At this point, all questions are sent to the user and all the answers returned by the user are obtained.

[0038] If the user selects answers 10, 20, and 30 for questions 1, 2, and 5 respectively, and the table cell to the right of question 3 is empty (meaning there are no new questions), then the user is informed that the product category is product category 1. Similarly, if the user selects answers 14, 24, 34, 40, and 50 for questions 1 through 5 respectively, and the table cell to the right of question 5 is empty, then the user is informed that the product category is product category 5.

[0039] The same judgment is made for the others to determine the product's category.

[0040] This embodiment dynamically loads a question-answer database, so modifying the answer to a question or changing the dialogue logic (i.e., the question order) only requires modifying the corresponding data in the database, without changing the code. Furthermore, using tables to store data occupies less storage space than storing data in a tree structure, and is easier to maintain and modify without requiring additional learning costs.

[0041] like Figure 2 As shown, the present invention also relates to a chatbot 100, comprising,

[0042] Memory 101 is used to store program instructions;

[0043] The processor 102 is configured to run the program instructions to perform the steps of the product identification and classification method based on human-computer interaction.

[0044] The embodiments disclosed in this specification are merely illustrative of one aspect of the invention, and the scope of protection of the invention is not limited to these embodiments. Any other functionally equivalent embodiments fall within the scope of protection of the invention. Those skilled in the art can make various other corresponding changes and modifications based on the technical solutions and concepts described above, and all such changes and modifications should fall within the scope of protection of the claims of this invention.

Claims

1. A product identification and classification method based on human-computer interaction, characterized in that, Includes the following steps: Step 1: Load the configuration file, which contains a database of product category relationships; Step 2: Load the first question and its corresponding answer options from the database in the configuration file and send them to the user, then receive the answer options returned by the user; Step 3: Continue sending the user the next question and its corresponding answer options until the current user's answer does not have a corresponding next question. Based on all the answers returned by the user, the system matches them against product categories in the database according to product classification relationships. If a matching product category is found, the user is notified that a suitable product category has been found; otherwise, the user is notified that a suitable product category could not be found. In step 3, after receiving a unique answer from the user for the current question, the answer is matched against the answer options for the same question in the database. If a match is found, it is checked whether there is a next question under that answer option. If so, the next question is sent to the user, and all the answers returned by the user are matched against the product categories in the database based on product classification relationships. If not, then directly match all the existing returned answers with the product categories in the database based on the product classification relationships. In the database of the configuration file, the product classification relationship logic is stored in tables. Each cell in the top row of the table stores a question. Each cell in the first column of the second row corresponds to a product category. Each cell to the right of each product category stores an answer to each question. Answers for the same question in the same column constitute the answer options. Load the first question in the table, send the question and the answer options corresponding to the column containing the question to the user, and return an answer from the answer options. After receiving the answer returned by the user, move the next table cell to the right from the row containing the answer. If the next table cell is not empty, move the column upwards to find the first question in the column containing the current answer, and send the found question and its corresponding answer options to the user again. Continue until the table cell to the right of the found answer is empty or does not exist. Then, send all questions to the user and obtain all the answers returned by the user.

2. The product identification and classification method based on human-computer interaction according to claim 1, characterized in that, After step 1 and before step 2, the following is also included: Obtain user data, predict the shortest classification path problem in the database based on the user data, and take the shortest classification path problem as the first problem.

3. A chatbot, characterized in that, include, Memory, used to store program instructions; A processor for running the program instructions to perform the steps of the product identification and classification method based on human-computer interaction as described in any one of claims 1-2.

Images