Precise commodity recommendation method and system based on data analysis

Abstract

The invention relates to the technical field of commodity recommendation, in particular to a commodity accurate recommendation method and system based on data analysis. The method comprises the following steps: acquiring historical browsing data of a user, dividing an operation threshold, identifying a mistaken touch behavior in the browsing data, and outputting mistaken touch behavior data; screening out the mistaken touch behavior data, obtaining target behavior data, constructing a user portrait, and carrying out commodity recommendation; detecting current browsing data of a user, dynamically updating a user portrait, obtaining updated commodity recommendation data, and outputting the updated commodity recommendation data; the system comprises a mistaken touch behavior recognition module, a user portrait construction module and a user portrait updating module. According to the embodiment of the invention, whether the error touch behavior exists in the browsing data is identified, the target behavior data is screened out, and the commodity recommendation is performed for the user according to the target behavior data, thereby improving the commodity recommendation precision.

Description

Technical Field

[0001] This invention relates to the field of product recommendation technology, and in particular to a method and system for accurate product recommendation based on data analysis. Background Technology

[0002] With the rapid development of technologies such as the Internet of Things (IoT) and big data, various technologies related to IoT and big data have been deeply applied to all aspects of social life. Currently, in online shopping, to improve user experience, relevant products are often recommended to users.

[0003] Existing methods make recommendations based on user browsing data, but when a user accidentally clicks on and exits the product browsing page, the same or similar products will still be pushed to the user, resulting in poor product recommendation accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for accurate product recommendation based on data analysis, which aims to solve the technical problem in the prior art that when a user accidentally clicks and exits product browsing, the same or similar products will still be pushed to the user, resulting in poor product recommendation accuracy.

[0005] To achieve the above objectives, the present invention employs a data analysis-based method for precise product recommendation, comprising the following steps:

[0006] Acquire user browsing history data, define operation thresholds, identify accidental touch behaviors in the browsing data, and output accidental touch behavior data;

[0007] Filter out accidental touch behavior data, obtain target behavior data, build user profiles, and make product recommendations;

[0008] Detect the user's current browsing data, dynamically update the user profile, obtain updated product recommendation data, and output the results.

[0009] Among the steps of acquiring user's historical browsing data, dividing operation thresholds, identifying accidental touch behaviors in the browsing data, and outputting accidental touch behavior data:

[0010] Obtain the user's historical browsing data and divide it into data fields. For each data field, query the dwell time and scrolling speed.

[0011] Output data fields indicating accidental touches based on dwell time and swipe speed;

[0012] Assign an identification stamp to the accidental touch behavior data and output the identification stamp and data fields.

[0013] In the step of outputting data fields for accidental touches based on dwell time and swipe speed:

[0014] Define a dwell time threshold to determine if a data field is accidentally touched. If the dwell time is less than the threshold, it is considered an accidental touch.

[0015] In the step of outputting data fields for accidental touches based on dwell time and swipe speed:

[0016] A speed threshold is defined to determine if a data field is accidentally touched. If the swiping speed is greater than the speed threshold, it is determined to be an accidental touch.

[0017] Among the steps of filtering out accidental touch behavior data, obtaining target behavior data, building user profiles, and making product recommendations:

[0018] Acquire valid behaviors, which are then used as target behavior data;

[0019] Calculate user preference weights for product categories based on target behavior data to generate user profiles;

[0020] The product recommendation score is calculated based on the user profile and product similarity.

[0021] After the step of calculating the product recommendation score based on user profile and product similarity:

[0022] Products are sorted based on recommendation scores, and then recommended to users in that order.

[0023] In the step of acquiring valid behavior, which is used as target behavior data:

[0024] Remove accidental touch data from historical browsing data, retain valid behaviors, and use these valid behaviors as target behavior data.

[0025] Among the steps, the following steps are involved: detecting the user's current browsing data, dynamically updating the user profile, obtaining updated product recommendation data, and outputting the results:

[0026] Real-time collection of user's current actions;

[0027] Update the user profile based on the current operation;

[0028] The recommendation score is recalculated based on the updated user profile, and a real-time recommendation list is output.

[0029] After collecting the user's current operation steps in real time:

[0030] Filter invalid data from the user's current operation data.

[0031] This invention also provides a data analysis-based accurate product recommendation system, including a mis-touch behavior recognition module, a user profile construction module, and a user profile update module; wherein:

[0032] The accidental touch behavior recognition module is used to acquire the user's historical browsing data, divide the operation threshold, identify accidental touch behaviors in the browsing data, and output accidental touch behavior data.

[0033] The user profile building module is used to filter out accidental touch behavior data, obtain target behavior data, build user profiles, and make product recommendations.

[0034] The user profile update module is used to detect the user's current browsing data, dynamically update the user profile, obtain updated product recommendation data, and output it.

[0035] This invention discloses a data analysis-based method and system for accurate product recommendation. The method comprises three modules: a mis-touch behavior identification module, a user profile construction module, and a user profile update module. The steps include: acquiring historical browsing data of the user, defining operation thresholds, identifying mis-touch behaviors in the browsing data, and outputting mis-touch behavior data; filtering out mis-touch behavior data, acquiring target behavior data, constructing a user profile, and recommending products; detecting the user's current browsing data, dynamically updating the user profile, acquiring updated product recommendation data, and outputting it; and improving product recommendation accuracy by identifying whether there are mis-touch behaviors in the browsing data, filtering out target behavior data, and recommending products to the user based on the target behavior data. Attached Figure Description

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 This is a flowchart of the steps of the data analysis-based accurate product recommendation method of the present invention.

[0038] Figure 2 This is a flowchart of steps S100 of the present invention.

[0039] Figure 3 This is a flowchart of steps S200 of the present invention.

[0040] Figure 4 This is a flowchart of steps S300 of the present invention.

[0041] Figure 5 This is a schematic diagram of the structural principle of the data analysis-based accurate product recommendation system of the present invention.

[0042] Figure 6 This is a schematic diagram of the electronic device of the present invention.

[0043] 401 - Accidental Touch Behavior Recognition Module, 402 - User Profile Construction Module, 403 - User Profile Update Module. Detailed Implementation

[0044] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0045] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0046] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0047] Please see Figures 1-4 This invention provides a method for accurate product recommendation based on data analysis, comprising the following steps:

[0048] S100: Acquire user's historical browsing data, divide operation thresholds, identify accidental touch behaviors in the browsing data, and output accidental touch behavior data.

[0049] In this embodiment, user browsing history data is acquired, operation thresholds are defined, accidental touch behaviors in the browsing data are identified, and accidental touch behavior data is output. The specific process is as follows:

[0050] S101: Obtain the user's historical browsing data and divide it into data fields. For each data field, query the dwell time and scrolling speed.

[0051] S102: Output data field indicating accidental touches based on dwell time and swipe speed;

[0052] S103: Assign an identification stamp to the accidental touch behavior data and output the identification stamp and data fields.

[0053] In the above process, user browsing history data is obtained, which can be extracted from a log system or database. The data fields include:

[0054] User ID (user_id)

[0055] Product ID (product_id)

[0056] Operation type (e.g., click, swipe, browse)

[0057] Dwell time (in seconds)

[0058] Swipe speed (in pixels per second)

[0059] Timestamp

[0060] Extract key fields (such as user_id, product_id, dwell_time, swipe_speed) from each record and store them as structured data (such as JSON or a database table).

[0061] Threshold T for determining dwell time threshold Thresholds can be dynamically set based on product type, for example:

[0062] Clothing: T threshold = 3 seconds;

[0063] Digital category: T threshold =5 seconds;

[0064] General formula:

[0065] T threshold (c) = β.avg_dwell_time(c);

[0066] Where β is an adjustment coefficient (e.g., 0.8); avg_dwell_time(c) is the average dwell time of product category c.

[0067] Threshold V for dividing sliding speed threshold Calculated based on device screen size:

[0068]

[0069] Among them, S screen γ is the screen pixel width (e.g., 1080 pixels); γ is an empirical coefficient (e.g., 1.5, indicating that the normal scrolling speed of the user is about 1.5 times the screen width per second).

[0070] The logic for judging accidental touch behavior can adopt the following rules:

[0071] Rule 1: If the dwell time is t <T threshold If so, it is determined to be an accidental touch.

[0072] Rule 2: If the sliding speed v > V threshold If so, it is determined to be an accidental touch.

[0073] Combined rule: If both rule 1 and rule 2 are satisfied, it is marked as a high-confidence false trigger.

[0074] Output accidental click information, generating an accidental click label (is_misclick, boolean value) for each data field:

[0075]

[0076] Assign an identification stamp to generate a unique identifier misclick_id for the accidental touch behavior data, and attach a timestamp and accidental touch type (such as "swipe too fast" or "stay too short").

[0077] The tagged accidental touch data is stored in a database or log file for subsequent analysis or filtering.

[0078] Example as follows:

[0079] Scenario: User A browses mobile products (screen width 1080 pixels), stays for 2 seconds, and scrolls at a speed of 900 pixels per second.

[0080] The dwell time threshold, assuming an average dwell time of 6 seconds for mobile devices, is as follows:

[0081] (T threshold (c) = β.avg_dwell_time(c) = 4.8 seconds.

[0082] User dwell time (t=2) seconds < 4.8 seconds → trigger rule 1.

[0083] Sliding speed threshold:

[0084]

[0085] User swiping speed (v = 900 pixels / second) > 720 pixels / second → trigger rule 2.

[0086] Accidental touch detection:

[0087] If both Rule 1 and Rule 2 are satisfied, it is marked as a high-confidence false trigger.

[0088] Output result:

[0089]

[0090] S200: Filter out accidental touch behavior data, obtain target behavior data, build user profiles, and make product recommendations.

[0091] In this implementation, accidental touch behavior data is filtered out, target behavior data is obtained, a user profile is constructed, and product recommendations are made. The specific process is as follows:

[0092] S201: Remove accidental touch behavior data from historical browsing data, retain valid behavior, and use the valid behavior as target behavior data;

[0093] S202: Calculate the user's preference weight for product categories based on target behavior data to generate a user profile;

[0094] S203: Calculate the product recommendation score based on user profile and product similarity;

[0095] S204: Sort the products according to the recommendation score and recommend products to the user in the sorted order.

[0096] During the above process, data filtering is performed to remove all data marked as accidental clicks (is_misclick=1) from the historical browsing data, while retaining valid behaviors (such as clicks, adding to cart, purchasing, and in-depth browsing).

[0097] Input: Raw historical data + accidental touch marker data (from step S100).

[0098] Output: Target behavior dataset, containing only valid operations.

[0099] Building user profiles based on preference weights allows for the assignment of weights to behaviors, such as:

[0100] Click: b click =1

[0101] Browsing (duration ≥ duration threshold): b browse =2

[0102] Add to cart: b cart =3

[0103] Purchase: b purchase =5

[0104] Time decay factor:

[0105] Recent user behavior should be given higher weight, as shown in the formula:

[0106] Decay(t) = e -λ·Δt ;

[0107] Where λ is the attenuation coefficient (e.g., 0.1); Δt is the difference between the time the behavior occurred and the current time (unit: days).

[0108] Preference weight calculation:

[0109]

[0110] Among them, b i The weight of the i-th action; t i Let be the timestamp of the i-th action.

[0111] User profile output generates a user-category preference matrix, for example:

[0112] User ID: U001; Category: Mobile Phone; Preference Weight: 1.875.

[0113] User ID: U001; Category: Headphones; Preference weight: 0.6.

[0114] Calculate the product recommendation score S u,p :

[0115] Calculate the similarity between product p and user preference category c based on collaborative filtering or content features (such as category, tag):

[0116]

[0117] Wherein, co-occurrence frequency (p,c) is the number of times product p is associated with category c; frequency (p) is the total number of times product p is exposed; and frequency (c) is the total number of products in category c.

[0118] Recommended score formula:

[0119]

[0120] Where C u A set of preference categories in a user profile.

[0121] Sort by recommendation score

[0122] Calculate S for all goods u,p And sorted in descending order.

[0123] Select the Top-K items (e.g., K=10) as the recommendation results.

[0124] Sample output:

[0125] Product ID: P002; Product Name: Smartphone X; Recommendation Score: 4.2.

[0126] Product ID: P003; Product Name: Wireless Earphones Y; Recommendation Score: 3.1.

[0127] Example as follows:

[0128] Scenario: User A's preference weight: mobile phone (W 手机 =1.875), headphones (W) 耳机 =0.6).

[0129] Product similarity: Similarity between mobile phone X and the category "mobile phone": Sim(X, mobile phone) = 0.9;

[0130] Similarity between headphone Y and the "headphone" category: Sim(Y, headphone) = 0.8.

[0131] Recommended score for Phone X:

[0132] S A,X =W 手机 Sim(X, mobile phone) = 1.875 × 0.9 = 1.6875.

[0133] Recommended score for headphones Y:

[0134] S A,X =W 耳机 Sim(Y, headphone) = 0.6 × 0.8 = 0.48.

[0135] Recommendation result: Mobile phone X (score 1.6875) is preferred over headphones Y (score 0.48).

[0136] S300: Detects the user's current browsing data, dynamically updates the user profile, obtains updated product recommendation data, and outputs it.

[0137] In this implementation, the user's current browsing data is detected, the user profile is dynamically updated, updated product recommendation data is obtained, and then output. The specific process is as follows:

[0138] S301: Real-time collection of user's current operations and filtering of invalid data;

[0139] S302: Update the user profile based on the current operation;

[0140] S303: Recalculate the recommendation score based on the updated user profile and output a real-time recommendation list.

[0141] During the above process, user behavior (such as clicks, swipes, pauses, adding to cart, purchasing, etc.) is captured in real time, and real-time behavior logs are generated.

[0142] As shown in step S100, the data field and invalid data filtering are performed by calling the accidental touch detection logic in step S100 to filter accidental touch behaviors, such as too short a dwell time or too fast a swipe speed.

[0143] Output: Real-time valid actions dataset.

[0144] Update user profile based on current operation.

[0145] Dynamically update preference weights W c Incremental calculation: Based on the existing user profile, preference weights are updated by combining real-time behavioral data; Formula:

[0146]

[0147] in, The old weight for category c in the user profile; The updated weights; α is a smoothing factor (e.g., 0.9), controlling the fusion ratio of the old and new weights; b new Weights for real-time behaviors (e.g., click = 1, purchase = 5); Decay(t new ) represents the time decay factor for real-time behavior; ∑ i∈batch Decay(t i $ represents the total time decay of the current batch of actions (if it is a single action, the denominator is 1).

[0148] The following is an example of a user profile update:

[0149] Original image:

[0150] User ID: U001; Category: Mobile Phone; Preference Weight: 1.875.

[0151] User ID: U001; Category: Headphones; Preference weight: 0.6.

[0152] New behavior: The user clicked on "Mobile X" (category = mobile phone), behavior weight b = 1, time decay factor ≈ 1.

[0153] Updated weights:

[0154]

[0155] Updated portrait:

[0156] User ID: U001; Category: Mobile Phone; Preference Weight: 1.7875.

[0157] User ID: U001; Category: Headphones; Preference weight: 0.6.

[0158] Recalculate the recommendation score based on the updated user profile and output a real-time recommendation list.

[0159] The recommendation score calculation calls the recommendation score formula in step S203, but uses the updated user profile weights.

[0160]

[0161] The product similarity calculation method in step S200 is used to obtain Sim(p,c).

[0162] Calculate for all candidate items Sort by descending order and select the Top-K items as the recommendation results.

[0163] The output example is as follows:

[0164] Updated user profile: Mobile phones (1.7875), Headphones (0.6).

[0165] Product similarity:

[0166] Mobile phone X (Sim(X, mobile phone) = 0.9);

[0167] The headphone value is Y(Sim(Y, headphone)=0.8).

[0168] Recommended score:

[0169] Mobile Phone X: 1.7875 × 0.9 = 1.60875;

[0170] Headphone Y: 0.6 × 0.8 = 0.48.

[0171] Recommended list:

[0172] User ID: P001; Category: Smartphone X; Preference weight: 1.60875.

[0173] User ID: P001; Category: Wireless Headphones Y; Preference weight: 0.48.

[0174] Corresponding to the aforementioned embodiments of the data analysis-based precise product recommendation method, this application also provides embodiments of a data analysis-based precise product recommendation system.

[0175] Figure 5 This is a block diagram illustrating a data analysis-based accurate product recommendation system according to an exemplary embodiment. (Refer to...) Figure 5 The system may include: a mis-touch behavior recognition module 401, a user profile construction module 402, and a user profile update module 403; wherein:

[0176] The accidental touch behavior recognition module 401 is used to acquire the user's historical browsing data, divide the operation threshold, identify accidental touch behaviors in the browsing data, and output accidental touch behavior data.

[0177] The user profile building module 402 is used to filter out accidental touch behavior data, obtain target behavior data, build user profiles, and make product recommendations.

[0178] The user profile update module 403 is used to detect the user's current browsing data, dynamically update the user profile, obtain updated product recommendation data, and output it.

[0179] In this embodiment, the accidental touch behavior recognition module 401 acquires the user's historical browsing data, divides operation thresholds, identifies accidental touch behaviors in the browsing data, and outputs accidental touch behavior data; the user profile construction module 402 filters out accidental touch behavior data, acquires target behavior data, constructs a user profile, and makes product recommendations; the user profile update module 403 detects the user's current browsing data, dynamically updates the user profile, acquires updated product recommendation data, and outputs it; by identifying whether there are accidental touch behaviors in the browsing data, and then filtering out target behavior data, product recommendations are made for the user based on the target behavior data, thereby improving the accuracy of product recommendations.

[0180] Regarding the system in the above embodiments, the specific ways in which each module performs operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0181] For the system embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this application according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0182] Accordingly, this application also provides an electronic device, including: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the data analysis-based accurate product recommendation method described above. Figure 6 The diagram shown is a hardware structure diagram of any device with data processing capabilities, used in an embodiment of the present invention to provide a precise product recommendation system based on data analysis. (Except for...) Figure 6 In addition to the processor, memory, and network interface shown, any data processing device in the embodiment may also include other hardware depending on the actual function of the data processing device, which will not be described in detail here.

[0183] Accordingly, this application also provides a computer-readable storage medium storing computer instructions thereon, which, when executed by a processor, implement the data analysis-based accurate product recommendation method described above. The computer-readable storage medium can be an internal storage unit of any data-processing device as described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data-processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data-processing device, and can also be used to temporarily store data that has been output or will be output.

[0184] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

[0185] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for accurate product recommendation based on data analysis, characterized in that, The steps include: Acquire user browsing history data, define operation thresholds, identify accidental touch behaviors in the browsing data, and output accidental touch behavior data; Filter out accidental touch behavior data, obtain target behavior data, build user profiles, and make product recommendations; Detect the user's current browsing data, dynamically update the user profile, obtain updated product recommendation data, and output the results.

2. The method for accurate product recommendation based on data analysis as described in claim 1, characterized in that, In the steps of acquiring user browsing history data, dividing operation thresholds, identifying accidental touch behaviors in the browsing data, and outputting accidental touch behavior data: Obtain the user's historical browsing data and divide it into data fields. For each data field, query the dwell time and scrolling speed. Output data fields indicating accidental touches based on dwell time and swipe speed; Assign an identification stamp to the accidental touch behavior data and output the identification stamp and data fields.

3. The data analysis-based accurate product recommendation method as described in claim 2, characterized in that, In the step of outputting data fields for accidental touches based on dwell time and swipe speed: Define a dwell time threshold to determine if a data field is accidentally touched. If the dwell time is less than the threshold, it is considered an accidental touch.

4. The data analysis-based precise product recommendation method as described in claim 3, characterized in that, In the step of outputting data fields for accidental touches based on dwell time and swipe speed: A speed threshold is defined to determine if a data field is accidentally touched. If the swiping speed is greater than the speed threshold, it is determined to be an accidental touch.

5. The method for accurate product recommendation based on data analysis as described in claim 1, characterized in that, In the steps of filtering out accidental touch behavior data, obtaining target behavior data, building user profiles, and making product recommendations: Acquire valid behaviors, which are then used as target behavior data; Calculate user preference weights for product categories based on target behavior data to generate user profiles; The product recommendation score is calculated based on the user profile and product similarity.

6. The method for accurate product recommendation based on data analysis as described in claim 5, characterized in that, After calculating the product recommendation score based on user profile and product similarity: Products are sorted based on recommendation scores, and then recommended to users in that order.

7. The method for accurate product recommendation based on data analysis as described in claim 6, characterized in that, In the step of acquiring valid behavior, which is used as target behavior data: Remove accidental touch data from historical browsing data, retain valid behaviors, and use these valid behaviors as target behavior data.

8. The method for accurate product recommendation based on data analysis as described in claim 7, characterized in that, In the steps of detecting the user's current browsing data, dynamically updating the user profile, obtaining updated product recommendation data, and outputting the results: Real-time collection of user's current actions; Update the user profile based on the current operation; The recommendation score is recalculated based on the updated user profile, and a real-time recommendation list is output.

9. The method for accurate product recommendation based on data analysis as described in claim 8, characterized in that, After collecting the user's current action steps in real time: Filter invalid data from the user's current operation data.

10. A data analysis-based product recommendation system, applied to the data analysis-based product recommendation method as described in claim 1, characterized in that, It includes a module for identifying accidental touches, a module for building user profiles, and a module for updating user profiles; among which: The accidental touch behavior recognition module is used to acquire the user's historical browsing data, divide the operation threshold, identify accidental touch behaviors in the browsing data, and output accidental touch behavior data. The user profile building module is used to filter out accidental touch behavior data, obtain target behavior data, build user profiles, and make product recommendations. The user profile update module is used to detect the user's current browsing data, dynamically update the user profile, obtain updated product recommendation data, and output it.

Images