Intelligent retrieval system based on artificial intelligence

Abstract

The invention discloses an intelligent retrieval method based on artificial intelligence, and the method comprises the steps: receiving retrieval information inputted by a user in a retrieval box, obtaining information data of historical shopping commodities and shopping carts / collection lists of the user in real time, and obtaining query habits of the user on a shopping platform as historical data of the user; constructing a user shopping feature preference model and a user shopping habit model based on the collected user historical data, performing semantic analysis on the retrieval information through an artificial intelligence technology, understanding the retrieval intention of the user, analyzing the commodity feature value required by the user, and determining the commodity feature value required by the user based on the user shopping habit model and the retrieval requirement of the user. The retrieval results are dynamically sorted and displayed on the current retrieval page, the sorting weights of the subsequent commodities are adjusted according to the clicking and staying conditions of the user on the retrieval result page, and the problems that personalized recommendation is single and the retrieval results are not intelligent when the user retrieves shopping commodities are solved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent retrieval technology, and in particular to an intelligent retrieval system based on artificial intelligence. Background Art

[0002] Intelligent retrieval system plays a vital role in shopping platforms. It helps users quickly find the products they need through efficient information retrieval technology, narrowing the user's search scope. The intelligent retrieval system of existing shopping platforms can already analyze user preferences for products, but the personalized recommendations for users are still relatively simple. Most of them are based on user preferences for product type, style, and price range. There are many repeated products in the search list, and users can only understand the differences in products after clicking in the search list, which wastes the user's time cost of searching for products and reduces shopping efficiency. In addition, when processing retrieval information, the intelligent retrieval system still uses the traditional method of extracting product keywords, and deletes all the modified words that are irrelevant to the description of the product. It often ignores the words that express the user's intention and cannot identify the user's specific needs for products, resulting in unintelligent retrieval results that cannot meet the user's more precise retrieval requirements. Users still need to filter products by manual browsing and judgment, which greatly reduces the user's overall shopping experience. Therefore, it is necessary to design an intelligent retrieval system based on artificial intelligence to improve personalized recommendations for shopping retrieval and provide more intelligent retrieval results. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent retrieval system based on artificial intelligence to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: an intelligent retrieval system based on artificial intelligence, which is executed in an intelligent retrieval method based on artificial intelligence, and the operation steps of the method include:

[0005] Step S1: receiving the search information entered by the user in the search box, obtaining the user's delivery address information on the shopping platform and the user's current IP address, and obtaining the user's historical shopping items and shopping cart / favorite list information data and the user's query habits on the shopping platform in real time as user history data;

[0006] Step S2: constructing a user shopping feature preference model and a user shopping habit model based on the collected user history data;

[0007] Step S3: semantically analyzing the search information through artificial intelligence technology, understanding the user's search intention, and analyzing the commodity feature values ​​required by the user;

[0008] Step S4: dynamically sorting the search results based on the user shopping habit model and the user's search needs, and displaying them on the current search page;

[0009] Step S5: According to the user's clicks and stays on the search result page, the ranking weights of subsequent products are adjusted.

[0010] According to the above technical solution, step S1 further includes the following steps:

[0011] Step S11: while receiving the search information input by the user in the search box, obtain the user's current IP address and time;

[0012] Step S12: the product information data of the historical shopping products and the shopping cart / favorite list includes: brand data, price data, product review quality data when purchasing the items, interaction time, and other data, wherein the other data is parameterized based on different characteristics of different products;

[0013] Step S13: The query habit includes: the number of clicks of the user in the search list and the query habit of the user after entering the product details page, and the query habit includes: checking comments, store ratings, and sales volume;

[0014] Step S14: add a shipping address index to the products on the shopping platform, analyze the shipping status of each merchant based on the merchant's historical shipping information, and add a shipping time node index and a shipping courier type index for the merchant.

[0015] According to the above technical solution, step S2 further includes the following steps:

[0016] Step S21: Based on the information data of the user's historical shopping products and shopping cart / favorite list, the K-means algorithm is used to classify different products, and the information data of the same category of products is analyzed for characteristics. The weights of the price data, the interaction time and the other data are relatively high, wherein the interaction time has a greater weight as the interaction time is closer. The brand data and the review quality data of the product at the time of purchase are used to calculate the user's preference for the i-th product feature of the o-category product, and to construct a user product preference model:

[0017] t oi =(price oi +review oi )×exp(buy oi +other oi +time oi )

[0018]

[0019] In the formula: o represents the type of product, i represents a product feature in the oth type of product, t oi represents the user's preference for the i-th product feature of category o, m represents the number of product types, n represents the number of product features under each type, and T represents the number of elements. oi The user product preference matrix, price oi represents the weight of the user's price of the feature of the i-th product in category o, review oi represents the weight of the user's evaluation quality of the i-th product feature of category o, buy oi represents the weight of the user's purchase of the i-th product feature of category o, other oi represents the weight of the user on the other features of the i-th product of category o, time oi represents the weight of the user's interaction time with the i-th product feature of category o;

[0020] Step S22: Based on the analysis of the user's historical data, the user's preference values ​​for different types of commodities are obtained, and the analysis results are aggregated and analyzed. The characteristics of the different types of commodities with different commodity characteristics are aggregated and analyzed to obtain the user's preference values ​​for different commodity characteristics in this type of commodity, wherein the calculation formula is:

[0021]

[0022]

[0023] Where t' oi represents the normalized feature preference value of the user for the i-th product feature of category o, t o Represents the preference value of product type o aggregating all product features i;

[0024] Step S23: Based on the analysis results of the historical data, a user feature preference model for different types of goods is constructed, and the feature preference model is updated in real time according to the user's operation on the shopping platform. The user's requirements for goods / merchants are analyzed according to the query habits. After the user enters the product details page, the user queries the product details page. When the user slides to the evaluation module on the product details page, if the number of evaluators is small or there are no comments, the product details page is directly exited. When the system recognizes that the situation exceeds the set threshold number of times, the system will ignore the search results for products with no evaluation / too few evaluations, and will not display them on the search page. The store rating and the sales volume also follow this step.

[0025] According to the above technical solution, step S3 further includes the following steps:

[0026] Step S31: Based on the search information input by the user in the current search box, the components included in the search information input by the user are analyzed, and the components included in the search information include: product keywords, irrelevant words, and user demand words. The user demand words refer to words that are not directly related to the product but can express the user's needs, such as: location, time, urgent need, XX express, many reviews, etc. The irrelevant words refer to words that have no actual reference significance for product retrieval, such as: modal particles, interjections or other expressions irrelevant to the search purpose;

[0027] Step S32: Based on the component analysis of the current search information, when the components of the current search information contain the user demand words, the user's intention is analyzed. When the user demand words are: urgent, arrive tomorrow, etc., which indicate that the user wants to receive the goods earlier, the user's current IP address and the user's delivery address in the shopping platform are obtained;

[0028] Step S33: When the user demand words are words such as XX express, location, and evaluation, the weights of these words are increased and classified as important keywords, and set as necessary search indexes.

[0029] According to the above technical solution, step S4 further includes the following steps:

[0030] Step S41: Based on the component analysis of the current search information, when the current search information component only includes the user demand words, the type of the currently searched product is determined according to the user shopping feature preference model and the user shopping habit model analyzed in steps S21-S23, and the user shopping feature preference corresponding to the product type is retrieved. According to the search results of the current search information, the products that meet the feature preference values ​​are sorted in descending order and displayed on the current search page;

[0031] Step S42: When the search information component analysis finds the user demand word in step S32, the distance between the product that meets the user's search and the address is calculated in a manner that the weight of the user's current IP address and the default delivery address is much greater than the weight of other delivery addresses. Based on the delivery time rule of the product, calculate the product link that can be delivered to the user as soon as possible and meets the requirements, and sort them in descending order according to the interval time from the user placing an order to receiving the goods. When the interval time difference between multiple products is less than the threshold set by the search system, the multiple products are sorted according to step S41 and displayed on the current search page;

[0032] Step S43: When the search information component is analyzed to the user demand word in step S32, the products that match the current search information will be further searched for products that match the user demand word, and the products that match the user demand word will be displayed first. The multiple products that match the user demand word are sorted according to step S41 and displayed on the current search page. Among them, when the user demand word is a location, after giving priority to displaying products that match the location, products that do not match the location are displayed first and then products that are close to the location.

[0033] According to the above technical solution, step S5 further includes the following steps:

[0034] Step S51: Based on the search result list displayed on the search page, according to the order of the products clicked by the user in the search result list, the user's browsing time is recorded, the part of the details page of the product that the user viewed is marked, and the part of the details page of the product that the user exited from is marked, and the reasons for the user's non-purchase are analyzed according to the part viewed by the user. According to the order in which the user views the parts, the system determines that the two parts viewed later are the reasons for the user's non-purchase, among which the weight of the last viewed part is much greater than the second to last viewed part, and the part viewed by the user first is determined by the system to meet the user's needs;

[0035] Step S52: Based on the user's behavior of browsing the search result list, the list that is not displayed later is adjusted, the similar products are compared with the product clicked by the user, the latter two parts are adjusted, and the first viewed part is retained for matching;

[0036] Step S53: Based on the search result list displayed on the search page, the user's characteristic preference for this type of product is updated according to the products clicked by the user, the products added to the shopping cart, and the time spent on the detail page.

[0037] According to the above technical solution, the system includes a retrieval information acquisition module, a shopping history analysis module, a retrieval information analysis module, and a real-time update feedback module:

[0038] The search information acquisition module is used to obtain the search information entered by the user on the search page, as well as the user's user history information in the shopping platform and the user's current IP address;

[0039] The shopping history analysis module is used to analyze user history data to build a user shopping feature preference model and a user shopping habit model;

[0040] The search information analysis module is used to analyze the search information input by the user to find out the user's shopping needs;

[0041] The real-time update feedback module is used to analyze the user's preference for commodities based on the user's click browsing operation on the search result page, and to update the subsequent search list sorting and the user's characteristic preference in real time.

[0042] According to the above technical solution, the shopping history analysis module includes a shopping feature preference module and a user shopping habit module:

[0043] The shopping feature preference module is used to analyze the information data of the historical shopping products and the shopping cart / favorite list, analyze the user's preference for different types of product features, and build a user shopping feature preference model;

[0044] The user shopping habit module is used to analyze the user's viewing habits on the product details page during the browsing process, analyze the user's requirements for the viewed content, and build a user shopping habit module.

[0045] According to the above technical solution, the retrieval information analysis module includes a commodity keyword module, an irrelevant word module, and a user demand word module:

[0046] The commodity keyword module is used to divide the current search information and divide the words describing the commodity into commodity keywords;

[0047] The irrelevant word module is used to classify the current search information and classify words that have no actual reference significance for product search as irrelevant words;

[0048] The user demand word module is used to classify the current search information and classify words that are not directly related to commodities but can express user demand into user demand words.

[0049] According to the above technical solution, the real-time update feedback module includes a user operation analysis module, a search result adjustment module, and an update feature preference module:

[0050] The user operation analysis module is used to record the clicks and browsing of the user in the search list results, the time of stay, the part of the product details page viewed, and the last part of the user exiting the product details page for subsequent reference;

[0051] The retrieval result adjustment module is used to match the products in the undisplayed list with the results of the analysis by the user operation analysis module and adjust the order;

[0052] The feature preference updating module is used to update the user's feature preferences for the products that the user clicks to browse, the products that are added to the shopping cart, and the time the user stays on the detail page.

[0053] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention provides an intelligent retrieval system for a shopping platform, which is classified according to the user's historical data in the platform, and on the basis of the existing retrieval system's recommendation based on the user's product preference analysis, it also adds considerations such as the user's evaluation quality, evaluation quantity, and product price of the product and the merchant, analyzes the user's query habits and requirements after entering the product details page, constructs a user shopping feature preference model and a user shopping habit model, and when analyzing user retrieval information, retains words that are irrelevant to the user's product description but can express the user's specific needs, which can further narrow the user's search scope, thereby improving the personalized recommendation of shopping retrieval and providing more intelligent retrieval results, thereby optimizing the user's shopping experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0055] Figure 1 A schematic diagram of the module composition of an artificial intelligence-based intelligent retrieval system provided in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] Embodiment 1: This embodiment can be applied to the scenario of XXX. The method can be executed by an intelligent retrieval system based on artificial intelligence provided by this embodiment. The method specifically includes the following steps:

[0058] Step S1: receiving the search information entered by the user in the search box, obtaining the user's delivery address information on the shopping platform and the user's current IP address, and obtaining the user's historical shopping items and shopping cart / favorite list information data and the user's query habits on the shopping platform in real time as user history data;

[0059] Step S2: constructing a user shopping feature preference model and a user shopping habit model based on the collected user history data;

[0060] Step S3: semantically analyzing the search information through artificial intelligence technology, understanding the user's search intention, and analyzing the commodity feature values ​​required by the user;

[0061] Step S4: dynamically sorting the search results based on the user shopping habit model and the user's search needs, and displaying them on the current search page;

[0062] Step S5: According to the user's clicks and stays on the search result page, the ranking weights of subsequent products are adjusted.

[0063] In an embodiment of the present invention, the step S1 further includes the following steps:

[0064] Step S11: while receiving the search information input by the user in the search box, obtain the user's current IP address and time;

[0065] Step S12: the product information data of the historical shopping products and the shopping cart / favorite list includes: brand data, price data, product review quality data when purchasing the items, interaction time, and other data, wherein the other data is parameterized based on different characteristics of different products;

[0066] Step S13: The query habit includes: the number of clicks of the user in the search list and the query habit of the user after entering the product details page, and the query habit includes: checking comments, store ratings, and sales volume;

[0067] Step S14: add a shipping address index to the products on the shopping platform, analyze the shipping status of each merchant based on the merchant's historical shipping information, and add a shipping time node index and a shipping courier type index for the merchant.

[0068] In an embodiment of the present invention, the step S2 further includes the following steps:

[0069] Step S21: Based on the information data of the user's historical shopping products and shopping cart / favorite list, the K-means algorithm is used to classify different products, and the information data of the same category of products is analyzed for characteristics. The weights of the price data, the interaction time and the other data are relatively high, wherein the interaction time has a greater weight as the interaction time is closer. The brand data and the review quality data of the product at the time of purchase are used to calculate the user's preference for the i-th product feature of the o-category product, and to construct a user product preference model:

[0070] t oi =(price oi +review oi )×exp(buy oi +other oi +time oi )

[0071]

[0072] In the formula: o represents the type of product, i represents a product feature in the oth type of product, t oi represents the user's preference for the i-th product feature of category o, m represents the number of product types, n represents the number of product features under each type, and T represents the number of elements. oi The user product preference matrix, price oi represents the weight of the user's price of the feature of the i-th product in category o, review oi represents the weight of the user's evaluation quality of the i-th product feature of category o, buy oi represents the weight of the user's purchase of the i-th product feature of category o, other oi represents the weight of the user on the other features of the i-th product of category o, time oi represents the weight of the user's interaction time with the i-th product feature of category o;

[0073] Step S22: Based on the analysis of the user's historical data, the user's preference values ​​for different types of commodities are obtained, and the analysis results are aggregated and analyzed. The characteristics of the different types of commodities with different commodity characteristics are aggregated and analyzed to obtain the user's preference values ​​for different commodity characteristics in this type of commodity, wherein the calculation formula is:

[0074]

[0075]

[0076] Where t' oi represents the normalized feature preference value of the user for the i-th product feature of category o, t o Represents the preference value of product type o aggregating all product features i;

[0077] Step S23: Based on the analysis results of the historical data, a user feature preference model for different types of goods is constructed, and the feature preference model is updated in real time according to the user's operation on the shopping platform. The user's requirements for goods / merchants are analyzed according to the query habits. After the user enters the product details page, the user queries the product details page. When the user slides to the evaluation module on the product details page, if the number of evaluators is small or there are no comments, the product details page is directly exited. When the system recognizes that the situation exceeds the set threshold number of times, the system will ignore the search results for products with no evaluation / too few evaluations, and will not display them on the search page. The store rating and the sales volume also follow this step.

[0078] In an embodiment of the present invention, step S3 further includes the following steps:

[0079] Step S31: Based on the search information input by the user in the current search box, the components included in the search information input by the user are analyzed, and the components included in the search information include: product keywords, irrelevant words, and user demand words. The user demand words refer to words that are not directly related to the product but can express the user's needs, such as: location, time, urgent need, XX express, many reviews, etc. The irrelevant words refer to words that have no actual reference significance for product retrieval, such as: modal particles, interjections or other expressions irrelevant to the search purpose;

[0080] Step S32: Based on the component analysis of the current search information, when the components of the current search information contain the user demand words, the user's intention is analyzed. When the user demand words are: urgent, arrive tomorrow, etc., which indicate that the user wants to receive the goods earlier, the user's current IP address and the user's delivery address in the shopping platform are obtained;

[0081] Step S33: When the user demand words are words such as XX express, location, and evaluation, the weights of these words are increased and classified as important keywords, and set as necessary search indexes.

[0082] In an embodiment of the present invention, the step S4 further includes the following steps:

[0083] Step S41: Based on the component analysis of the current search information, when the current search information component only includes the user demand words, the type of the currently searched product is determined according to the user shopping feature preference model and the user shopping habit model analyzed in steps S21-S23, and the user shopping feature preference corresponding to the product type is retrieved. According to the search results of the current search information, the products that meet the feature preference values ​​are sorted in descending order and displayed on the current search page;

[0084] Step S42: When the search information component analysis finds the user demand word in step S32, the distance between the product that meets the user's search and the address is calculated in a manner that the weight of the user's current IP address and the default delivery address is much greater than the weight of other delivery addresses. Based on the delivery time rule of the product, calculate the product link that can be delivered to the user as soon as possible and meets the requirements, and sort them in descending order according to the interval time from the user placing an order to receiving the goods. When the interval time difference between multiple products is less than the threshold set by the search system, the multiple products are sorted according to step S41 and displayed on the current search page;

[0085] Step S43: When the search information component is analyzed to the user demand word in step S32, the products that match the current search information will be further searched for products that match the user demand word, and the products that match the user demand word will be displayed first. The multiple products that match the user demand word are sorted according to step S41 and displayed on the current search page. Among them, when the user demand word is a location, after giving priority to displaying products that match the location, products that do not match the location are displayed first and then products that are close to the location.

[0086] In an embodiment of the present invention, the step S5 further includes the following steps:

[0087] Step S51: Based on the search result list displayed on the search page, according to the order of the products clicked by the user in the search result list, the user's browsing time is recorded, the part of the details page of the product that the user viewed is marked, and the part of the details page of the product that the user exited from is marked, and the reasons for the user's non-purchase are analyzed according to the part viewed by the user. According to the order in which the user views the parts, the system determines that the two parts viewed later are the reasons for the user's non-purchase, among which the weight of the last viewed part is much greater than the second to last viewed part, and the part viewed by the user first is determined by the system to meet the user's needs;

[0088] Step S52: Based on the user's behavior of browsing the search result list, the list that is not displayed later is adjusted, the similar products are compared with the product clicked by the user, the latter two parts are adjusted, and the first viewed part is retained for matching;

[0089] Step S53: Based on the search result list displayed on the search page, the user's characteristic preference for this type of product is updated according to the products clicked by the user, the products added to the shopping cart, and the time spent on the detail page.

[0090] Embodiment 2: Embodiment 2 of the present invention provides an intelligent retrieval system based on artificial intelligence. Figure 1 A schematic diagram of the module composition of an intelligent retrieval system based on artificial intelligence provided in the second embodiment of the present invention is shown in FIG. Figure 1 As shown, the system includes a retrieval information acquisition module, a shopping history analysis module, a retrieval information analysis module, and a real-time update feedback module:

[0091] The search information acquisition module is used to obtain the search information entered by the user on the search page, as well as the user's user history information in the shopping platform and the user's current IP address;

[0092] The shopping history analysis module is used to analyze user history data to build a user shopping feature preference model and a user shopping habit model;

[0093] The search information analysis module is used to analyze the search information input by the user to find out the user's shopping needs;

[0094] The real-time update feedback module is used to analyze the user's preference for commodities based on the user's click browsing operation on the search result page, and to update the subsequent search list sorting and the user's characteristic preference in real time.

[0095] In some embodiments of the present invention, the shopping history analysis module includes a shopping feature preference module and a user shopping habit module:

[0096] The shopping feature preference module is used to analyze the information data of the historical shopping products and the shopping cart / favorite list, analyze the user's preference for different types of product features, and build a user shopping feature preference model;

[0097] The user shopping habit module is used to analyze the user's viewing habits on the product details page during the browsing process, analyze the user's requirements for the viewed content, and build a user shopping habit module.

[0098] In some embodiments of the present invention, the retrieval information analysis module includes a commodity keyword module, an irrelevant word module, and a user demand word module:

[0099] The commodity keyword module is used to divide the current search information and divide the words describing the commodity into commodity keywords;

[0100] The irrelevant word module is used to classify the current search information and classify words that have no actual reference significance for product search as irrelevant words;

[0101] The user demand word module is used to classify the current search information and classify words that are not directly related to commodities but can express user demand into user demand words.

[0102] In some embodiments of the present invention, the real-time update feedback module includes a user operation analysis module, a search result adjustment module, and a feature preference update module:

[0103] The user operation analysis module is used to record the clicks and browsing of the user in the search list results, the time of stay, the part of the product details page viewed, and the last part of the user exiting the product details page for subsequent reference;

[0104] The retrieval result adjustment module is used to match the products in the undisplayed list with the results of the analysis by the user operation analysis module and adjust the order;

[0105] The feature preference updating module is used to update the user's feature preferences for the products that the user clicks to browse, the products that are added to the shopping cart, and the time the user stays on the detail page.

[0106] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0107] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent retrieval method based on artificial intelligence, The method comprises the following steps: Step S1: receiving the search information entered by the user in the search box, obtaining the user's delivery address information on the shopping platform and the user's current IP address, and obtaining the user's historical shopping items and shopping cart / favorite list information data and the user's query habits on the shopping platform in real time as user history data; Step S2: Building a user shopping feature preference model and a user shopping habit model based on the collected user history data; Step S3: semantically analyzing the search information through artificial intelligence technology to understand the user's search intent and analyze the commodity feature values ​​required by the user; Step S4: dynamically sorting the search results based on the user shopping habit model and the user's search needs, and displaying them on the current search page; Step S5: According to the user's clicks and stays on the search result page, the ranking weights of subsequent products are adjusted.

2. The intelligent retrieval method based on artificial intelligence according to claim 1, characterized in that: The step S1 further comprises the following steps: Step S11: while receiving the search information input by the user in the search box, obtain the user's current IP address and time; Step S12: the product information data of the historical shopping products and the shopping cart / favorite list includes: brand data, price data, product review quality data when purchasing the items, interaction time, and other data, wherein the other data is parameterized based on different characteristics of different products; Step S13: The query habit includes: the number of clicks of the user in the search list and the query habit of the user after entering the product details page, and the query habit includes: checking comments, store ratings, and sales volume; Step S14: add a shipping address index to the products on the shopping platform, analyze the shipping status of each merchant based on the merchant's historical shipping information, and add a shipping time node index and a shipping courier type index for the merchant.

3. The intelligent retrieval method based on artificial intelligence according to claim 2, characterized in that: The step S2 further comprises the following steps: Step S21: Based on the information data of the user's historical shopping products and shopping cart / favorite list, the K-means algorithm is used to classify different products, and the information data of the same category of products is analyzed for characteristics. The weights of the price data, the interaction time and the other data are relatively high, wherein the interaction time has a greater weight as the interaction time is closer. The brand data and the review quality data of the product at the time of purchase are used to calculate the user's preference for the i-th product feature of the o-category product, and to construct a user product preference model: t oi =(price oi +review oi )×exp(buy oi +other oi +time oi ) In the formula: o represents the type of product, i represents a product feature in the oth type of product, t oi represents the user's preference for the i-th product feature of category o, m represents the number of product types, n represents the number of product features under each type, and T represents the element t oi The user product preference matrix, price oi represents the weight of the user's price of the feature of the i-th product in category o, review oi represents the weight of the user's evaluation quality of the i-th product feature of category o, buy i represents the weight of the user's purchase of the i-th product feature of category o, other oi represents the weight of the user on the other features of the i-th product of category o, time oi represents the weight of the user's interaction time with the i-th product feature of category o; Step S22: Based on the analysis of the user's historical data, the user's preference values ​​for different types of commodities are obtained, and the analysis results are aggregated and analyzed. The characteristics of the different types of commodities with different commodity characteristics are aggregated and analyzed to obtain the user's preference values ​​for different commodity characteristics in this type of commodity, wherein the calculation formula is: In the formula, t' oi represents the normalized feature preference value of the user for the i-th product feature of category o, t o Represents the preference value of product type o aggregating all product features i; Step S23: Based on the analysis results of the historical data, a user feature preference model for different types of goods is constructed, and the feature preference model is updated in real time according to the user's operation on the shopping platform. The user's requirements for goods / merchants are analyzed according to the query habits. After the user enters the product details page, the user queries the product details page. When the user slides to the evaluation module on the product details page, if the number of evaluators is small or there are no comments, the product details page is directly exited. When the system recognizes that the situation exceeds the set threshold number of times, the system will ignore the search results for products with no evaluation / too few evaluations, and will not display them on the search page. The store rating and the sales volume also follow this step.

4. The intelligent retrieval method based on artificial intelligence according to claim 3, characterized in that: The step S3 further comprises the following steps: Step S31: Based on the search information input by the user in the current search box, the components included in the search information input by the user are analyzed, wherein the components included in the search information include: product keywords, irrelevant words, and user demand words, wherein the user demand words refer to words that are not directly related to the product but can express the user's needs, and the irrelevant words refer to words that have no actual reference significance for product search; Step S32: Based on the component analysis of the current search information, when the components of the current search information contain the user demand words, the user's intention is analyzed. When the user demand words are: urgent, arrive tomorrow, etc., which indicate that the user wants to receive the goods earlier, the user's current IP address and the user's delivery address in the shopping platform are obtained; Step S33: When the user demand words are words such as XX express, location, and evaluation, the weights of these words are increased and classified as important keywords, and set as necessary search indexes.

5. The intelligent retrieval method based on artificial intelligence according to claim 4, characterized in that: The step S4 further comprises the following steps: Step S41: Based on the component analysis of the current search information, when the current search information component only includes the user demand words, the type of the currently searched product is determined according to the user shopping feature preference model and the user shopping habit model analyzed in steps S21-S23, and the user shopping feature preference corresponding to the product type is retrieved. According to the search results of the current search information, the products that meet the feature preference values ​​are sorted in descending order and displayed on the current search page; Step S42: When the search information component analysis finds the user demand word in step S32, the distance between the product that meets the user's search and the address is calculated in a manner that the weight of the user's current IP address and the default delivery address is much greater than the weight of other delivery addresses. Based on the delivery time rule of the product, calculate the product link that can be delivered to the user as soon as possible and meets the requirements, and sort them in descending order according to the interval time from the user placing an order to receiving the goods. When the interval time difference between multiple products is less than the threshold set by the search system, the multiple products are sorted according to step S41 and displayed on the current search page; Step S43: When the search information component is analyzed to the user demand word in step S32, the products that match the current search information will be further searched for products that match the user demand word, and the products that match the user demand word will be displayed first. The multiple products that match the user demand word are sorted according to step S41 and displayed on the current search page. Among them, when the user demand word is a location, after giving priority to displaying products that match the location, products that do not match the location are displayed first and then products that are close to the location.

6. The intelligent retrieval method based on artificial intelligence according to claim 5, characterized in that: The step S5 further comprises the following steps: Step S51: Based on the search result list displayed on the search page, according to the order of the products clicked by the user in the search result list, the user's browsing time is recorded, the part of the details page of the product that the user viewed is marked, and the part of the details page of the product that the user exited from is marked, and the reasons for the user's non-purchase are analyzed according to the part viewed by the user. According to the order in which the user views the parts, the system determines that the two parts viewed later are the reasons for the user's non-purchase, among which the weight of the last viewed part is much greater than the second to last viewed part, and the part viewed by the user first is determined by the system to meet the user's needs; Step S52: Based on the user's behavior of browsing the search result list, the list that is not displayed later is adjusted, the similar products are compared with the product clicked by the user, the latter two parts are adjusted, and the first viewed part is retained for matching; Step S53: Based on the search result list displayed on the search page, the user's characteristic preference for this type of product is updated according to the products clicked by the user, the products added to the shopping cart, and the time spent on the detail page.

7. An intelligent retrieval system based on artificial intelligence, characterized in that: The system includes a retrieval information acquisition module, a shopping history analysis module, a retrieval information analysis module, and a real-time update feedback module: The search information acquisition module is used to obtain the search information entered by the user on the search page, as well as the user's user history information in the shopping platform and the user's current IP address; The shopping history analysis module is used to analyze user history data to build a user shopping feature preference model and a user shopping habit model; The search information analysis module is used to analyze the search information input by the user to find out the user's shopping needs; The real-time update feedback module is used to analyze the user's preference for commodities based on the user's click browsing operation on the search result page, and to update the subsequent search list sorting and the user's characteristic preference in real time.

8. The intelligent retrieval system based on artificial intelligence according to claim 7, characterized in that: The shopping history analysis module includes a shopping feature preference module and a user shopping habit module: The shopping feature preference module is used to analyze the information data of the historical shopping products and the shopping cart / favorite list, analyze the user's preference for different types of product features, and build a user shopping feature preference model; The user shopping habit module is used to analyze the user's viewing habits on the product details page during the browsing process, analyze the user's requirements for the viewed content, and build a user shopping habit module.

9. The intelligent retrieval system based on artificial intelligence according to claim 8, characterized in that: The retrieval information analysis module includes a commodity keyword module, an irrelevant word module, and a user demand word module: The commodity keyword module is used to divide the current search information and divide the words describing the commodity into commodity keywords; The irrelevant word module is used to classify the current search information and classify words that have no actual reference significance for product search as irrelevant words; The user demand word module is used to classify the current search information and classify words that are not directly related to commodities but can express user demand into user demand words.

10. The intelligent retrieval system based on artificial intelligence according to claim 9, characterized in that: The real-time update feedback module includes a user operation analysis module, a search result adjustment module, and a feature preference update module: The user operation analysis module is used to record the clicks and browsing of the user in the search list results, the time of stay, the part of the product details page viewed, and the last part of the user exiting the product details page for subsequent reference; The retrieval result adjustment module is used to match the products in the undisplayed list with the results of the analysis by the user operation analysis module and adjust the order; The feature preference updating module is used to update the user's feature preferences for the products that the user clicks to browse, the products that are added to the shopping cart, and the time the user stays on the detail page.

Images