A customer profiling method based on knowledge graph

Abstract

The present invention relates to the field of data processing technology, and in particular to a customer portrait method based on a knowledge graph, which determines the number of nodes for constructing a blank data structure based on the amount of historical browsing data; obtains customer input data and determines a comprehensive evaluation value; arranges the data in the data set and places them on corresponding nodes to generate a knowledge graph under the condition that the data set meets a preset standard; based on the knowledge graph, inputs target data to extract feature data, and determines whether the accuracy of the customer portrait of the knowledge graph meets the preset standard based on the similarity comparison result between the feature data and the data set; and determines adjustment parameters based on the difference in portrait accuracy under the condition that the accuracy of the customer portrait does not meet the preset standard. The present invention improves the accuracy of customer portraits based on knowledge graphs.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a customer profiling method based on knowledge graph. Background Art

[0002] As a data structure that represents entities and relationships, knowledge graphs can efficiently integrate and represent massive amounts of heterogeneous information, providing strong support for computers to understand and process natural language texts. In the field of customer profiling, knowledge graphs make it possible to comprehensively and accurately portray customers by constructing a complex network of relationships between customer entities and various attributes, behaviors, preferences, etc. This method uses advanced entity recognition, relationship extraction, entity linking, and entity embedding technologies to extract and integrate customer information from multi-source data to form a rich customer profiling label system. These labels not only cover the basic attributes of customers, but also deeply reveal their interests, behavior patterns, and potential needs, providing a solid data foundation for enterprises' precision marketing, personalized recommendations, and intelligent decision-making. Therefore, the customer profiling method based on knowledge graphs has become an important application direction of big data and artificial intelligence technologies in the field of customer analysis.

[0003] Chinese patent application publication number: CN118093865A discloses a customer profiling method and system based on knowledge graph, which obtains the customer's historical travel records and the post-ride evaluation corresponding to each historical travel record; determines the customer's preferred itinerary based on the historical travel records; determines the customer's route knowledge graph based on the customer's preferred itinerary; performs text analysis on the post-ride evaluation to obtain the customer's customer portrait, which is the customer's emotional stance on the itinerary; determines the itinerary plan pushed to the customer and the customer preference information pushed to the car owner based on the route knowledge graph and the customer portrait. In this way, due to the addition of the customer portrait, that is, the customer's emotional stance on the itinerary, assisted by the route knowledge graph, the itinerary plan pushed to the customer is determined, thereby simplifying the customer's operation steps.

[0004] It can be seen that the customer profiling method and system based on the knowledge graph have the following problems: the invention pushes travel plans to customers based on the route knowledge graph and customer profile, but the invention lacks monitoring of the accuracy of the route knowledge graph, resulting in low accuracy of the customer profiling based on the route knowledge graph. Summary of the invention

[0005] To this end, the present invention provides a customer profiling method based on a knowledge graph to overcome the problem of lack of monitoring of the accuracy of the route knowledge graph in the prior art, thereby resulting in low accuracy of the customer profiling based on the knowledge graph.

[0006] To achieve the above object, the present invention provides a customer profiling method based on knowledge graph, comprising:

[0007] Determine the number of nodes for constructing a blank data structure based on a comparison result between the amount of historical browsing data of the customer and the preset amount of data;

[0008] Obtaining customer input data, obtaining first-category keywords based on the customer input data and determining first information completeness;

[0009] Determine the amount of associated data based on the historical browsing data associated with the first category of keywords, obtain the second category of keywords based on the amount of associated data and determine the second information completeness;

[0010] generating a data set based on the first category of keywords and the second category of keywords, determining a comprehensive evaluation value based on the first information completeness and the second information completeness, and determining whether the data set meets a preset standard based on the comprehensive evaluation value;

[0011] Under the condition that the data set meets the preset standard, the data in the data set are arranged and placed on corresponding nodes to generate a knowledge graph;

[0012] Based on the knowledge graph, target data is input to extract feature data, and based on the similarity comparison result between the feature data and the data set, whether the accuracy of the customer portrait of the knowledge graph meets the preset standard is determined;

[0013] Under the condition that the accuracy of the customer portrait does not meet the preset standard, determining to adjust the preset data volume based on the portrait accuracy difference, or adjusting the associated data volume;

[0014] Among them, the first category of keywords are factual keywords, and the second category of keywords are evaluative keywords.

[0015] Further, the process of determining the number of nodes of the blank data structure based on the amount of historical browsing data includes:

[0016] If the amount of historical browsing data is greater than or equal to the preset amount of data, determining to construct the blank data structure with the first number of nodes;

[0017] If the amount of historical browsing data is less than the preset amount of data, it is determined to construct the blank data structure with a second number of nodes.

[0018] Further, the process of determining the comprehensive evaluation value based on the first category of keywords and the second category of keywords includes:

[0019] Determining the first information completeness based on the overlap between the first category of keywords and the first category of preset keywords;

[0020] Determining the second information completeness based on the overlap between the second category keywords and the second category preset keywords;

[0021] The comprehensive evaluation value is determined based on the first information completeness and the second information completeness.

[0022] Furthermore, the process of determining whether the data set meets the preset standard based on the comprehensive evaluation value includes:

[0023] If the comprehensive evaluation value is greater than or equal to a preset comprehensive evaluation value, it is determined that the data set meets the preset standard;

[0024] If the comprehensive evaluation value is less than the preset comprehensive evaluation value, it is determined that the data set does not meet the preset standard;

[0025] The data set is a set of the first category of keywords and the second category of keywords.

[0026] Further, under the condition that the data set does not meet the preset standard, the process of determining to adjust the number of the first category of preset keywords based on the comprehensive evaluation value difference includes:

[0027] Subtracting a preset comprehensive evaluation value from the comprehensive evaluation value to obtain a comprehensive evaluation value difference;

[0028] Compare the comprehensive evaluation value difference with the preset evaluation value difference to obtain a comparison result;

[0029] Determining to increase the number of the first category of preset keywords based on a comparison result that the comprehensive evaluation value difference is greater than the preset evaluation value difference;

[0030] Therein, several adjustment methods are set for the number of the first category preset keywords, and each adjustment method has a different adjustment range for the number of the first category preset keywords.

[0031] Further, under the condition that the data set does not meet the preset standard, the process of determining to adjust the number of the second category preset keywords based on the comprehensive evaluation value difference includes:

[0032] Subtracting a preset comprehensive evaluation value from the comprehensive evaluation value to obtain a comprehensive evaluation value difference;

[0033] Compare the comprehensive evaluation value difference with the preset evaluation value difference to obtain a comparison result;

[0034] Determining to increase the number of the second category preset keywords based on the comparison result that the comprehensive evaluation value difference is less than or equal to the preset evaluation value difference;

[0035] Therein, several adjustment methods are set for the number of the second category preset keywords, and each adjustment method has a different adjustment range for the number of the second category preset keywords.

[0036] Furthermore, the process of generating a knowledge graph under the condition that the data set meets the preset standard includes:

[0037] For each data in the data set, the number of its labels is counted and the data is arranged in descending order based on the number of labels of each data, and recorded as the first data, the second data, ..., the nth data, respectively, where n is the total number of data in the data set;

[0038] Based on the number of connection relationships of each node in the blank data structure, the nodes are arranged in descending order based on the number of connection relationships of each node, and are recorded as the first node, the second node, and the mth node, respectively, where m is the total number of nodes in the blank data structure;

[0039] Placing the first data on the first node, placing the second data on the second node, and setting them sequentially to generate a knowledge graph;

[0040] Here, m is greater than or equal to n.

[0041] Furthermore, the process of inputting the target data based on the knowledge graph to determine whether the portrait accuracy of the knowledge graph meets the preset standard includes:

[0042] Inputting the target data to extract the feature data;

[0043] Determine data similarity based on comparing the feature data with the data set;

[0044] If the data similarity is greater than or equal to a preset data similarity, it is determined that the accuracy of the portrait meets the preset standard;

[0045] If the data similarity is less than the preset data similarity, it is determined that the portrait accuracy does not meet the preset standard.

[0046] Further, under the condition that the portrait accuracy does not meet the preset standard, the process of determining to adjust the preset data amount based on the portrait accuracy difference includes:

[0047] Comparing the portrait accuracy difference with a preset accuracy difference;

[0048] Determine the difference comparison result that the portrait accuracy difference is greater than the preset accuracy difference;

[0049] Determining to adjust the preset data amount based on the difference comparison result;

[0050] Among them, several adjustment methods are set for the preset data amount, and each adjustment method has a different adjustment range for the preset data amount.

[0051] Furthermore, the process of determining and adjusting the amount of associated data based on the portrait accuracy difference includes:

[0052] Comparing the portrait accuracy difference with a preset accuracy difference;

[0053] Determine a difference comparison result in which the portrait accuracy difference is less than or equal to the preset accuracy difference;

[0054] Determining to adjust the amount of associated data based on the difference comparison result;

[0055] Therein, several adjustment methods are set for the associated data amount, and each adjustment method has a different adjustment range for the associated data amount.

[0056] Compared with the prior art, the beneficial effects of the present invention lie in that the present invention determines the number of nodes for constructing a data structure based on the amount of historical browsing data, thereby improving the utilization efficiency of the data structure; determines whether a data set meets a preset standard based on a comprehensive evaluation value, thereby improving the judgment accuracy for the data set, and constructs a knowledge graph when the data set meets the preset standard, thereby improving the construction benchmark of the knowledge graph; verifies the portrait accuracy of the knowledge graph based on the target data, and determines the adjustment parameters based on the portrait accuracy difference, thereby improving the accuracy of the customer portrait based on the knowledge graph.

[0057] Furthermore, the present invention determines the number of nodes for constructing a data structure based on the amount of historical browsing data. A large number of nodes is set for a large amount of historical browsing data, and a small number of nodes is set for a small amount of historical browsing data. The more nodes there are, the more accurate the output result will be, and the data processing speed will correspondingly become slower. The number of nodes is determined based on the amount of historical browsing data, which improves the accuracy of customer portraits based on the knowledge graph while ensuring the data processing speed.

[0058] Furthermore, the present invention determines whether a data set meets preset standards based on a comprehensive evaluation value, thereby improving the judgment standard of the data set, and when the data set does not meet the preset standards, determines to increase the number of first-category preset keywords or second-category preset keywords based on the difference in the comprehensive evaluation value, and sets several adjustment ranges for the number of preset keywords based on the difference, thereby accurately improving the judgment benchmark and further improving the accuracy of customer portraits based on knowledge graphs.

[0059] Furthermore, the present invention arranges the data based on the number of labels of each data in the data set, places the data with the largest number of labels on the nodes with the most connection relationships, and arranges them in sequence, thereby improving the analysis accuracy of the knowledge graph, thereby further improving the accuracy of the customer portrait based on the knowledge graph.

[0060] Furthermore, the present invention verifies the portrait accuracy of the knowledge graph based on the target data. The knowledge graph extracts feature data based on the target data and outputs the results. The feature data is keywords, which aims to eliminate unnecessary data and irrelevant data based on the search results of the browser, thereby further improving the accuracy of the customer portrait based on the knowledge graph. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a flow chart of a customer profiling method based on knowledge graph according to an embodiment of the present invention;

[0062] Figure 2 A flowchart for determining the number of nodes of a blank data structure according to an embodiment of the present invention;

[0063] Figure 3 A flowchart of an embodiment of the present invention for determining whether the data set meets a preset standard;

[0064] Figure 4 A flowchart for determining whether the portrait accuracy of the knowledge graph meets preset standards in an embodiment of the present invention. DETAILED DESCRIPTION

[0065] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0066] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0067] See also Figure 1 As shown, it is a flow chart of a customer profiling method based on knowledge graph in an embodiment of the present invention.

[0068] The embodiment of the present invention provides a customer profiling method based on knowledge graph, including:

[0069] Step S1, determining the number of nodes for constructing a blank data structure based on a comparison result between the amount of historical browsing data of the customer and a preset amount of data;

[0070] Step S2, obtaining customer input data, obtaining first-category keywords based on the customer input data and determining first information completeness;

[0071] Step S3, determining the amount of associated data based on the historical browsing data associated with the first category of keywords, acquiring the second category of keywords based on the amount of associated data and determining the second information completeness;

[0072] Step S4, generating a data set based on the first category of keywords and the second category of keywords, determining a comprehensive evaluation value based on the first information completeness and the second information completeness, and determining whether the data set meets a preset standard based on the comprehensive evaluation value;

[0073] Step S5, under the condition that the data set meets the preset standard, the data in the data set are arranged and placed on corresponding nodes to generate a knowledge graph;

[0074] Step S6, based on the knowledge graph, input the target data to extract feature data, and determine whether the accuracy of the customer portrait of the knowledge graph meets the preset standard based on the similarity comparison result between the feature data and the data set;

[0075] Step S7, if the customer portrait accuracy does not meet the preset standard, determine to adjust the preset data volume based on the portrait accuracy difference, or adjust the associated data volume.

[0076] Specifically, the information completeness is determined based on the matching degree between the keywords extracted in real time and the preset keywords.

[0077] Specifically, the amount of associated data is positively correlated with the number of first-category keywords. The more the number of first-category keywords is, the larger the amount of associated data is.

[0078] Specifically, the first category of keywords are factual keywords such as name, number of people, address, and turnover, which are not specifically limited; the second category of keywords are evaluative keywords such as credit value and number of corporate risks, which are not specifically limited.

[0079] Specifically, the present invention determines the number of nodes for constructing a data structure based on the amount of historical browsing data, thereby improving the utilization efficiency of the data structure; determines whether a data set meets a preset standard based on a comprehensive evaluation value, thereby improving the judgment accuracy for the data set, and constructs a knowledge graph when the data set meets the preset standard, thereby improving the construction benchmark of the knowledge graph; verifies the portrait accuracy of the knowledge graph based on the target data, and determines the adjustment parameters based on the portrait accuracy difference, thereby improving the accuracy of the customer portrait based on the knowledge graph.

[0080] See also Figure 2As shown, it is a flow chart of determining the number of nodes of a blank data structure according to an embodiment of the present invention.

[0081] Specifically, the process of determining the number of nodes of the blank data structure based on the amount of historical browsing data includes:

[0082] If the amount of historical browsing data is greater than or equal to the preset amount of data, it is determined to construct the blank data structure with the first node quantity 30;

[0083] If the amount of historical browsing data is less than the preset amount of data, it is determined to construct the blank data structure with a second number of nodes 20.

[0084] Specifically, the preset data volume is the median value of the customer's historical browsing data volume.

[0085] Specifically, the present invention determines the number of nodes for constructing a data structure based on the amount of historical browsing data. A large number of nodes is set for a large amount of historical browsing data, and a small number of nodes is set for a small amount of historical browsing data. The more nodes there are, the more accurate the output result will be, and the data processing speed will correspondingly become slower. The number of nodes is determined based on the amount of historical browsing data, which improves the accuracy of customer portraits based on the knowledge graph while ensuring the data processing speed.

[0086] Specifically, the process of determining the comprehensive evaluation value based on the first category of keywords and the second category of keywords includes:

[0087] Determining the first information completeness based on the overlap between the first category of keywords and the first category of preset keywords;

[0088] Determining the second information completeness based on the overlap between the second category keywords and the second category preset keywords;

[0089] The comprehensive evaluation value is determined based on the first information completeness and the second information completeness.

[0090] Specifically, the keywords in the first category that are the same as the keywords in the first category of preset keywords are overlapped, the overlap degree is the ratio of the number of overlaps with the first category of preset keywords to the total number of the first category of preset keywords, and the information completeness is the overlap degree between the keywords and the preset keywords.

[0091] Specifically, the comprehensive evaluation value is the sum of the product of the first information completeness and its corresponding weight coefficient, and the product of the second information completeness and its corresponding weight coefficient.

[0092] Specifically, the weight coefficient is set according to the customer's preference, and the specific value is not limited.

[0093] See also Figure 3As shown, it is a flowchart of an embodiment of the present invention for determining whether the data set meets the preset standard.

[0094] Specifically, the process of determining whether the data set meets the preset standard based on the comprehensive evaluation value includes:

[0095] If the comprehensive evaluation value is greater than or equal to a preset comprehensive evaluation value of 0.7, it is determined that the data set meets the preset standard;

[0096] If the comprehensive evaluation value is less than the preset comprehensive evaluation value of 0.7, it is determined that the data set does not meet the preset standard;

[0097] The data set is a set of the first category of keywords and the second category of keywords.

[0098] Specifically, under the condition that the data set does not meet the preset standard, the process of determining to adjust the first category of preset keywords based on the comprehensive evaluation value difference includes:

[0099] Determine the comparison result that the comprehensive evaluation value difference is greater than the preset evaluation value difference of 0.03;

[0100] Determining to increase the number of the first category of preset keywords based on the comparison result;

[0101] Therein, several adjustment methods are set for the number of the first category preset keywords, and each adjustment method has a different adjustment range for the number of the first category preset keywords.

[0102] Specifically, the difference between the comprehensive evaluation value difference and the preset evaluation value difference is recorded as the first difference.

[0103] If the first difference is greater than the first preset difference of 0.01, it is determined to use the first keyword correction coefficient of 1.3 to correct the number of the first category of preset keywords;

[0104] If the first difference is less than or equal to the first preset difference of 1%, it is determined to use the second keyword correction coefficient of 1.1 to correct the number of the first category of preset keywords.

[0105] Specifically, under the condition that the data set does not meet the preset standard, the process of determining to adjust the second category of preset keywords based on the comprehensive evaluation value difference includes:

[0106] Determine the comparison result that the comprehensive evaluation value difference is less than or equal to the preset evaluation value difference;

[0107] Determining to increase the number of the second category of preset keywords based on the comparison result;

[0108] Therein, several adjustment methods are set for the number of the second category preset keywords, and each adjustment method has a different adjustment range for the number of the second category preset keywords.

[0109] Specifically, the difference between the preset evaluation value difference and the comprehensive evaluation value difference is recorded as the second difference;

[0110] If the second difference is greater than the second preset difference of 2%, it is determined to use the third keyword correction coefficient of 1.4 to correct the number of the second category preset keywords;

[0111] If the second difference is less than or equal to the second preset difference of 2%, it is determined to use the fourth keyword correction coefficient of 1.2 to correct the number of second category preset keywords.

[0112] Specifically, the present invention determines whether a data set meets preset standards based on a comprehensive evaluation value, improves the judgment standard of the data set, and determines to increase the number of first-category preset keywords or second-category preset keywords based on the difference in the comprehensive evaluation value when the data set does not meet the preset standards, and sets several adjustment ranges for the number of preset keywords based on the difference, thereby accurately improving the judgment benchmark and further improving the accuracy of customer portraits based on knowledge graphs.

[0113] Specifically, the process of generating a knowledge graph under the condition that the data set meets the preset standards includes:

[0114] For each data in the data set, the number of its labels is counted and the data is arranged in descending order based on the number of labels of each data, and recorded as the first data, the second data, ..., the nth data, respectively, where n is the total number of data in the data set;

[0115] Based on the number of connection relationships of each node in the blank data structure, and based on the number of connection relationships of each node, the nodes are arranged in descending order and recorded as a first node, a second node, ..., an mth node, respectively, where m is the total number of nodes in the blank data structure;

[0116] Placing the first data on the first node, placing the second data on the second node, and setting them sequentially to generate a knowledge graph;

[0117] Here, m is greater than or equal to n.

[0118] Specifically, the present invention arranges the data based on the number of labels of each data in the data set, places the data with the largest number of labels on the nodes with the most connection relationships, and arranges them in sequence, thereby improving the analysis accuracy of the knowledge graph, and further improving the accuracy of the customer portrait based on the knowledge graph.

[0119] See also Figure 4 As shown, it is a flowchart of an embodiment of the present invention for determining whether the portrait accuracy of the knowledge graph meets the preset standard.

[0120] Specifically, the process of inputting the target data based on the knowledge graph to determine whether the portrait accuracy of the knowledge graph meets the preset standard includes:

[0121] Inputting the target data to extract the feature data;

[0122] Determine data similarity based on comparing the feature data with the data set;

[0123] If the data similarity is greater than or equal to 10% of the preset data similarity, it is determined that the accuracy of the portrait meets the preset standard;

[0124] If the data similarity is less than 10% of the preset data similarity, it is determined that the portrait accuracy does not meet the preset standard.

[0125] Specifically, the feature data is the keywords related to customer needs extracted by the knowledge graph based on the target data, and the data similarity is the degree of overlap between the feature data extracted by the knowledge graph and the keywords in the data set.

[0126] Specifically, under the condition that the portrait accuracy does not meet the preset standard, the process of determining to adjust the preset data amount based on the portrait accuracy difference includes:

[0127] Determine the difference comparison result of the image accuracy difference greater than 10% of the preset accuracy difference;

[0128] Determining to adjust the preset data amount based on the difference comparison result;

[0129] Among them, several adjustment methods are set for the preset data amount, and each adjustment method has a different adjustment range for the preset data amount.

[0130] Specifically, the difference between the portrait accuracy difference and the preset accuracy difference is recorded as the third difference;

[0131] If the third difference is greater than the preset third difference of 5%, the preset data volume is adjusted by using the first data volume adjustment coefficient of 1.5;

[0132] If the third difference is less than or equal to the preset third difference of 5%, the preset data amount is adjusted using the second data amount adjustment coefficient of 1.3.

[0133] Specifically, the process of determining and adjusting the amount of associated data based on the difference in portrait accuracy includes:

[0134] Determine the difference comparison result with a portrait accuracy difference less than or equal to a preset accuracy difference of 1%;

[0135] Determining to adjust the amount of associated data based on the difference comparison result;

[0136] Therein, several adjustment methods are set for the associated data amount, and each adjustment method has a different adjustment range for the associated data amount.

[0137] Specifically, the difference between the preset accuracy difference and the portrait accuracy difference is recorded as the fourth difference;

[0138] If the fourth difference is greater than the preset fourth difference of 3%, the first data amount adjustment coefficient of 1.4 is used to adjust the associated data amount;

[0139] If the fourth difference is less than or equal to the preset fourth difference of 3%, the second data amount adjustment coefficient of 1.2 is used to adjust the associated data amount.

[0140] Specifically, the present invention verifies the portrait accuracy of the knowledge graph based on the target data, and determines to adjust the preset data amount or the associated data amount based on the portrait accuracy difference when the accuracy does not meet the preset standard, thereby accurately overcoming the problems arising in the knowledge graph construction process, thereby further improving the accuracy of the customer portrait based on the knowledge graph.

[0141] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A customer profiling method based on knowledge graph, characterized in that: include: Determine the number of nodes for constructing a blank data structure based on a comparison result between the amount of historical browsing data of the customer and the preset amount of data; Obtaining customer input data, obtaining first-category keywords based on the customer input data and determining first information completeness; Determine the amount of associated data based on the historical browsing data associated with the first category of keywords, obtain the second category of keywords based on the amount of associated data and determine the second information completeness; generating a data set based on the first category of keywords and the second category of keywords, determining a comprehensive evaluation value based on the first information completeness and the second information completeness, and determining whether the data set meets a preset standard based on the comprehensive evaluation value; Under the condition that the data set meets the preset standard, the data in the data set are arranged and placed on corresponding nodes to generate a knowledge graph; Based on the knowledge graph, target data is input to extract feature data, and based on the similarity comparison result between the feature data and the data set, it is determined whether the accuracy of the customer portrait of the knowledge graph meets the preset standard; Under the condition that the accuracy of the customer portrait does not meet the preset standard, determining to adjust the preset data volume based on the portrait accuracy difference, or adjusting the associated data volume; Among them, the first category of keywords are factual keywords, and the second category of keywords are evaluative keywords; The process of determining whether the data set meets the preset standard based on the comprehensive evaluation value includes: If the comprehensive evaluation value is greater than or equal to a preset comprehensive evaluation value, it is determined that the data set meets the preset standard; If the comprehensive evaluation value is less than the preset comprehensive evaluation value, it is determined that the data set does not meet the preset standard; Wherein, the data set is a set of the first category of keywords and the second category of keywords; Under the condition that the data set does not meet the preset standard, the process of determining to adjust the number of the first category preset keywords based on the comprehensive evaluation value difference includes: Subtracting a preset comprehensive evaluation value from the comprehensive evaluation value to obtain a comprehensive evaluation value difference; Compare the comprehensive evaluation value difference with the preset evaluation value difference to obtain a comparison result; Determining to increase the number of the first category of preset keywords based on a comparison result that the comprehensive evaluation value difference is greater than the preset evaluation value difference; Therein, several adjustment methods are set for the number of the first category preset keywords, and each adjustment method has a different adjustment range for the number of the first category preset keywords.

2. The customer profiling method based on knowledge graph according to claim 1 is characterized in that: The process of determining the number of nodes of the blank data structure based on the amount of historical browsing data includes: If the amount of historical browsing data is greater than or equal to the preset amount of data, determining to construct the blank data structure with the first number of nodes; If the amount of historical browsing data is less than the preset amount of data, it is determined to construct the blank data structure with a second number of nodes.

3. The customer profiling method based on knowledge graph according to claim 1 is characterized in that: The process of determining the comprehensive evaluation value based on the first category of keywords and the second category of keywords includes: Determining the first information completeness based on the overlap between the first category of keywords and the first category of preset keywords; Determining the second information completeness based on the overlap between the second category keywords and the second category preset keywords; The comprehensive evaluation value is determined based on the first information completeness and the second information completeness.

4. The customer profiling method based on knowledge graph according to claim 3 is characterized in that: Under the condition that the data set does not meet the preset standard, the process of determining to adjust the number of the second category preset keywords based on the comprehensive evaluation value difference includes: Subtracting a preset comprehensive evaluation value from the comprehensive evaluation value to obtain a comprehensive evaluation value difference; Compare the comprehensive evaluation value difference with the preset evaluation value difference to obtain a comparison result; Determining to increase the number of the second category preset keywords based on the comparison result that the comprehensive evaluation value difference is less than or equal to the preset evaluation value difference; Therein, several adjustment methods are set for the number of the second category preset keywords, and each adjustment method has a different adjustment range for the number of the second category preset keywords.

5. The customer profiling method based on knowledge graph according to claim 4 is characterized in that: The process of generating a knowledge graph under the condition that the data set meets the preset standards includes: For each data in the data set, the number of its labels is counted and the data is arranged in descending order based on the number of labels of each data, and recorded as the first data, the second data, ..., the nth data, respectively, where n is the total number of data in the data set; Based on the number of connection relationships of each node in the blank data structure, the nodes are arranged in descending order based on the number of connection relationships of each node, and are recorded as the first node, the second node, and the mth node, respectively, where m is the total number of nodes in the blank data structure; Placing the first data on the first node, placing the second data on the second node, and setting them sequentially to generate a knowledge graph; Here, m is greater than or equal to n.

6. The customer profiling method based on knowledge graph according to claim 5 is characterized in that: The process of inputting the target data based on the knowledge graph to determine whether the portrait accuracy of the knowledge graph meets the preset standard includes: Inputting the target data to extract the feature data; Determine data similarity based on comparing the feature data with the data set; If the data similarity is greater than or equal to the preset data similarity, it is determined that the portrait accuracy of the knowledge graph meets the preset standard; If the data similarity is less than the preset data similarity, it is determined that the portrait accuracy of the knowledge graph does not meet the preset standard.

7. The customer profiling method based on knowledge graph according to claim 6 is characterized in that: Under the condition that the portrait accuracy does not meet the preset standard, the process of determining to adjust the preset data amount based on the portrait accuracy difference includes: Comparing the portrait accuracy difference with a preset accuracy difference; Determine the difference comparison result that the portrait accuracy difference is greater than the preset accuracy difference; Determining to adjust the preset data amount based on the difference comparison result; Among them, several adjustment methods are set for the preset data amount, and each adjustment method has a different adjustment range for the preset data amount.

8. The customer profiling method based on knowledge graph according to claim 7 is characterized in that: The process of determining and adjusting the amount of associated data based on the difference in portrait accuracy includes: Comparing the portrait accuracy difference with a preset accuracy difference; Determine a difference comparison result in which the portrait accuracy difference is less than or equal to the preset accuracy difference; Determining to adjust the amount of associated data based on the difference comparison result; Therein, several adjustment methods are set for the associated data amount, and each adjustment method has a different adjustment range for the associated data amount.

Images