Label portrait analysis method and device

Abstract

The invention discloses a label portrait analysis method and device, and relates to the technical field of computers. A specific embodiment of the method comprises the steps of obtaining a target entity set and a historical entity set corresponding to a target tag; generating a link ratio difference set of the target label according to the target entity set and the historical entity set; feature data of each tag is constructed, a stability result of each tag is obtained according to the feature data of each tag, and the tags comprise a target tag and a portrait tag corresponding to the target tag; and for any portrait label, under the condition that both the stability result of the portrait label and the stability result of the target label indicate stability, generating a target portrait result of the target label according to the link ratio difference set. According to the embodiment, rapid portrait analysis can be carried out on a single label, the timeliness of label portrait analysis is improved, resource consumption is reduced, and large-scale expansion of the number of portrait types can be realized.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a label image analysis method and device. Background Art

[0002] In the field of e-commerce, labeling systems for people, products, scenes, etc. are widely used. The quality of labels has an important impact on their application. Label image analysis can not only evaluate the quality of labels, but also serve as a reference for iterative label updates.

[0003] In related technologies, label portrait analysis mainly uses offline engines to perform entity field association matching to generate portrait results. However, this method has low time efficiency and requires association of the entire data each time, which consumes a lot of resources. It only supports analysis of a small number of labels and a small number of portrait types and cannot be flexibly expanded on a large scale. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides a label portrait analysis method and device, which can realize rapid portrait analysis of a single label, improve the timeliness of label portrait analysis, reduce resource consumption, and realize large-scale expansion of the number of portrait types.

[0005] To achieve the above object, according to one aspect of an embodiment of the present invention, a tag portrait analysis method is provided, comprising:

[0006] Get the target entity set and the historical entity set corresponding to the target label;

[0007] Generate a month-on-month difference set of the target label according to the target entity set and the historical entity set;

[0008] Constructing feature data of each tag, and obtaining a stability result of each tag according to the feature data of each tag, wherein the tags include a target tag and a portrait tag corresponding to the target tag;

[0009] For any portrait tag, when the stability result of the portrait tag and the stability result of the target tag both indicate stability, a target portrait result of the target tag is generated according to the month-on-month difference set.

[0010] Optionally, generating a month-on-month difference set of the target labels according to the target entity set and the historical entity set includes:

[0011] The target entity set and the historical entity set are subjected to difference calculation to obtain the month-on-month difference set, wherein the month-on-month difference set includes an incremental difference set and / or a decremental difference set, wherein the incremental difference set includes target entities that are not in the historical entity set, and the decremental difference set includes historical entities that are not in the target entity.

[0012] Optionally, performing difference calculation on the target entity set and the historical entity set includes:

[0013] Convert each target entity in the target entity set into a corresponding target entity code, and construct a target bitmap according to each target entity code;

[0014] Convert each historical entity in the historical entity set into a corresponding historical entity code, and construct a historical bitmap according to each historical entity code;

[0015] The target bitmap and the historical bitmap are subjected to difference calculation to generate an incremental bitmap of the incremental difference set and / or a decremental bitmap of the decremental difference set.

[0016] Optionally, obtaining a stability result of each tag according to the characteristic data includes:

[0017] Inputting feature data of the label into a classification model to obtain a stability result of the label, wherein the feature data includes one or more of a label source data feature, a label calculation rule feature, a label result data feature, and a label history portrait feature;

[0018] The classification model is obtained by the following method:

[0019] Obtain multiple sample labels and stability results for each sample label;

[0020] Construct feature data for each sample label;

[0021] The classification model is trained based on the feature data and stability results of each sample label.

[0022] Optionally, the month-on-month difference set includes an incremental difference set and a decremental difference set, and generating a target portrait result of the target tag according to the month-on-month difference set includes:

[0023] Calculate the incremental portrait result corresponding to the incremental difference set and the portrait label;

[0024] Calculate the reduction portrait result corresponding to the reduction difference set and the portrait label;

[0025] Obtain the historical portrait results corresponding to the historical entity set and the portrait label;

[0026] The target portrait result is generated according to the historical portrait result, the incremental portrait result and the reduced portrait result.

[0027] Optionally, calculating the incremental portrait result corresponding to the incremental difference set and the portrait label includes:

[0028] Obtain an entity set corresponding to the portrait label;

[0029] The incremental difference set is associated with the entity set to generate the incremental portrait result.

[0030] Optionally, after generating the target portrait result of the target tag according to the month-on-month difference set, the method further includes:

[0031] The quality warning information of the target label is generated according to the target profiling result.

[0032] According to another aspect of an embodiment of the present invention, there is provided a label image analysis device, comprising:

[0033] An acquisition module is used to acquire a target entity set and a historical entity set corresponding to a target label;

[0034] A generation module, generating a month-on-month difference set of the target label according to the target entity set and the historical entity set;

[0035] A determination module is used to construct feature data of each tag and obtain a stability result of each tag according to the feature data of each tag, wherein the tag includes a target tag and a portrait tag corresponding to the target tag;

[0036] The analysis module generates, for any portrait tag, a target portrait result of the target tag according to the month-on-month difference set when the stability result of the portrait tag and the stability result of the target tag both indicate stability.

[0037] According to another aspect of an embodiment of the present invention, there is provided an electronic device, including:

[0038] one or more processors;

[0039] a storage device for storing one or more programs,

[0040] When the one or more programs are executed by the one or more processors, the one or more processors implement the label portrait analysis method provided by the present invention.

[0041] According to another aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored, and when the program is executed by a processor, the label portrait analysis method provided by the present invention is implemented.

[0042] An embodiment of the above invention has the following advantages or beneficial effects: the label portrait analysis method of the embodiment of the present invention first obtains the target entity set and the historical entity set corresponding to the target label, and generates a month-on-month difference set based on the target entity set and the historical entity set; then, by constructing the characteristic data of each label, the stability result of each label is obtained; when the stability results of the target label and the portrait label both indicate stability, the target portrait result of the target label is obtained according to the month-on-month difference set. This method realizes rapid portrait analysis of a single label through the stability result and the month-on-month difference set of the label, and can realize real-time or near real-time label portrait analysis, thereby improving timeliness; label portrait analysis based on the month-on-month difference set does not require full data association, thereby reducing resource consumption; it supports the expansion of the number of portrait types, and can flexibly expand on a large scale.

[0043] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0045] Figure 1 is a schematic diagram of the main process of a tag portrait analysis method according to an embodiment of the present invention;

[0046] Figure 2 is a schematic diagram of the main process of another tag portrait analysis method according to an embodiment of the present invention;

[0047] Figure 3 is a schematic diagram of the main process of another tag portrait analysis method according to an embodiment of the present invention;

[0048] Figure 4 is a schematic diagram of main modules of a label image analysis device according to an embodiment of the present invention;

[0049] Figure 5 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0050] Figure 6 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0052] At present, labeling systems for people, goods, and scenes are widely used in the field of e-commerce, such as direct regularization of tagging people, selecting products, and selecting stores based on tags, or using tags as the basic feature data of marketing / risk control algorithm models, and performing classification, prediction, and evaluation through algorithms, thereby supporting applications such as crowd analysis, product marketing, and scene activity construction. The quality of tags is key information for tag application. By performing a portrait analysis on tags, the quality of tags can be evaluated and used as a reference for rapid iterative updates of tags. For example, an activity plans to promote laptop computers to college students and constructs a college freshman tag. The people marked by the tag do not meet the preset age distribution or gender ratio, indicating that there is a problem with the quality of the tag and it needs to be corrected and iterated.

[0053] In related technologies, label portrait analysis mainly uses offline engines such as Spark to perform entity field association matching to generate portrait results. This method has low efficiency and cannot generate results in real time or near real time. In addition, it associates the entire data each time, consumes a lot of resources, and can only support analysis of a small number of labels and a small number of portrait types, and cannot be flexibly expanded on a large scale.

[0054] Based on the above technical problems, an embodiment of the present invention provides a label portrait analysis method, which can realize rapid portrait analysis of a single label and support the expansion of the number of portrait types to achieve label quality assessment.

[0055] Figure 1 is a schematic diagram of the main process of a tag portrait analysis method according to an embodiment of the present invention. Figure 1 As shown, the label portrait analysis method includes the following steps:

[0056] Step S101: Acquire a target entity set and a historical entity set corresponding to a target tag;

[0057] Step S102: Generate a month-on-month difference set of target labels according to the target entity set and the historical entity set;

[0058] Step S103: constructing feature data of each tag, and obtaining the stability result of each tag according to the feature data of each tag, wherein the tags include a target tag and a portrait tag corresponding to the target tag;

[0059] Step S104: For any portrait tag, when the stability result of the portrait tag and the stability result of the target tag both indicate stability, a target portrait result of the target tag is generated according to the month-on-month difference set.

[0060] In an embodiment of the present invention, in the label portrait analysis method, firstly, a target entity set and a historical entity set corresponding to the target label are obtained, the target entity set is a set of entities marked with the target label in a first time period, and the historical entity set is a set of entities marked with the target label in a second time period. The entity is an identifier of a marked object, and the object may be a user, a commodity, a store, a scene, etc. If the entity is a user identifier, the user identifier may be a user account; the second time period is earlier than the first time period, and the second time period and the first time period may be equal in length, so that the month-on-month difference set may be calculated.

[0061] In an embodiment of the present invention, the label portrait analysis is to perform entity type portrait analysis on the target entity set marked by the target label. For example, the target entity set is a high-potential user set, a light luxury bag product set, etc. Performing entity type portrait analysis on the target entity set can be used to analyze the user interests and hobbies of the high-potential group, the product marketing strategy of light luxury products, etc.

[0062] In an embodiment of the present invention, after obtaining the target entity set and the historical entity set, a month-on-month difference set of the target label is generated according to the target entity set and the historical entity set. The month-on-month difference set represents the change of the target entity set compared with the historical entity set. The month-on-month difference set includes newly added entities and reduced entities, that is, the month-on-month difference set includes an incremental set consisting of newly added entities and a decremental set consisting of reduced entities.

[0063] In an embodiment of the present invention, generating a month-on-month difference set of target labels according to a target entity set and a historical entity set includes:

[0064] The target entity set and the historical entity set are subjected to difference calculation to obtain a month-on-month difference set, which includes an incremental difference set and / or a decremental difference set. The incremental difference set includes the target entities that are not in the historical entity set, and the decremental difference set includes the historical entities that are not in the target entity.

[0065] In an embodiment of the present invention, a difference set is calculated between a target entity set and a historical entity set, such as comparing each target entity in the target entity set with each historical entity in the historical entity set, to obtain an incremental difference set and / or a decremental difference set, wherein the incremental difference set is a set consisting of newly added target entities, and the decremental difference set is a set consisting of reduced historical entities. The incremental difference set or the decremental difference set may be an empty set. When both are empty sets, it indicates that the target label is relatively stable and the quality of the target label is high.

[0066] In an embodiment of the present invention, the target entity in the target entity set may be compared with the historical entity set in the historical entity set to obtain a month-on-month difference set, so as to analyze the target portrait result of the target tag using the month-on-month difference set.

[0067] In the embodiment of the present invention, Figure 2 As shown, the target entity set and the historical entity set are subjected to difference calculation, including:

[0068] Step S201: convert each target entity in the target entity set into a corresponding target entity code, and construct a target bitmap according to each target entity code;

[0069] Step S202: convert each historical entity in the historical entity set into a corresponding historical entity code, and construct a historical bitmap according to each historical entity code;

[0070] Step S203: Calculate the difference between the target bitmap and the historical bitmap to generate an incremental bitmap of the incremental difference set and / or a decremental bitmap of the decremental difference set.

[0071] In an embodiment of the present invention, when generating a month-on-month difference set of a target tag, first, each target entity in the target entity set can be converted into a target entity code according to a preset rule, and each historical entity in the historical entity set can be converted into a historical entity code, such as converting a user account into a corresponding user digital id according to a preset rule, the same entity corresponds to the same user digital id, and different entities correspond to different user digital ids. The preset rule can be customized, such as the preset rule can be a mapping relationship between a pre-configured entity and an id. Then, a target bitmap is constructed according to the set of target entity codes, and a historical bitmap is constructed according to the set of historical entity codes, that is, the target entity code and the historical entity code are both represented by bytes, and a target bitmap and a historical bitmap are obtained respectively. Then, the target bitmap and the historical bitmap are calculated for difference, and the difference calculation can be performed through a diff function to obtain an incremental bitmap and a decrement bitmap, the incremental bitmap corresponds to the incremental difference set, and the decrement bitmap corresponds to the decrement difference set, that is, a month-on-month difference set is obtained, wherein the incremental bitmap is the difference obtained by subtracting the historical bitmap from the target bitmap, and the decrement bitmap is the difference obtained by subtracting the incremental bitmap from the historical bitmap. By converting the target entity set and the historical entity set into bitmaps and using the bitmaps for difference calculations, the efficiency of generating target label difference sets can be improved, thereby improving the efficiency of label portrait analysis.

[0072] Among them, bitmap is an efficient integer set representation and set calculation method. Each integer in the set corresponds to a bit in the bitmap.

[0073] In an embodiment of the present invention, the stability result of each label is obtained according to the characteristic data of each label, including: inputting the characteristic data of the label into a classification model to obtain the stability result of the label, the characteristic data including one or more of label source data characteristics, label calculation rule characteristics, label result data characteristics, and label history portrait characteristics.

[0074] In an embodiment of the present invention, after generating a month-on-month difference set of the target label, the stability of the target label is evaluated. Label stability refers to the evaluation of the variability of the number of label results and the distribution of label values. Specifically, the feature data of each label is constructed, and the label includes a target label and a portrait label corresponding to the target label. The target label is a label to be subjected to portrait analysis, and the portrait label is a label used to perform portrait analysis on the target label. The target label and the portrait label are determined according to business needs. The portrait label can be one or more, and the portrait label corresponding to the target label can be customized according to business needs, or can be selected according to the set rules. For example, if the target label is a college freshman, it is necessary to conduct an age group portrait analysis on the college freshman, and the portrait label can be an age label for the college freshman. For each label, the feature data of each label is constructed, and the feature data of each label can be constructed according to the target entity set and the historical entity set. The feature data of each label may include one or more of the features of label source data, label calculation rule features, label result data features and label historical portrait features, wherein the label source data features are constructed based on the label source data information, such as the source table, source data time range, source table data fluctuation, etc. Different source tables, source data time ranges, source table data fluctuations, etc. can be processed by one-hot encoding to obtain the label source data features; the label calculation rule features are constructed based on the label calculation rule information, such as finding the intersection of table A and table B, calculating the maximum value, calculating the average value, etc. Different calculation rules can be processed by one-hot encoding, so as to obtain the label source data features. The label calculation rule characteristics are obtained; the label result data characteristics are constructed based on the label result data information, which may include information such as changes in the total amount of data and changes in the distribution of sampled data, such as how much the total amount of data corresponding to the label has increased or decreased. If the label result data information is numerical data, the numerical data can be normalized and then one-hot encoded to obtain the label result data characteristics; the label history portrait characteristics are constructed based on the core portrait change information in the label history, such as the change ratio of the product specification portrait distribution and other information. For example, the change ratio of the product specification portrait distribution can be normalized first, and then one-hot encoded to obtain the label history portrait characteristics.

[0075] A feature matrix is ​​constructed based on one or more of the features of the label source data, the features of the label calculation rule data, the features of the label result data, and the features of the label history portrait, so as to obtain the feature data of each label. Then, the feature data of the label is input into the trained classification model, and the classification model is used to classify the stability of the label to obtain the stability result of the label. The stability result can indicate stability or instability. Thus, the feature data of each label is input into the classification model, and the labels can be classified by the classification model to obtain the stability result of each label. The stability result indicates whether the label is a stable label or an unstable label, so as to evaluate the stability of the label. Among them, the stability result can be represented by a stability value, such as a stability value of [0~1], 0 means completely unstable, that is, the number of label results and the label value distribution corresponding to the target entity set and the historical entity set are completely different, and 1 means completely stable, that is, the number of label results and the label value distribution corresponding to the target entity set and the historical entity set are unchanged. After obtaining the stability result of each label, it is stored in the system to facilitate the subsequent determination of the portrait analysis process. The stability of the label is evaluated through the classification model to improve the accuracy and efficiency of the stability evaluation, and the rapid portrait analysis process is determined by determining the stability type of each label to achieve rapid portrait analysis of the label.

[0076] Among them, the classification model can evaluate and predict the stability of the label through reinforcement learning, neural network or other classification algorithms. When training the classification model, first obtain multiple sample labels and the stability results of each sample label, then build the feature data of each sample label, and train according to the feature data and stability results of each sample label to obtain the classification model. It is also possible to use reinforcement learning, neural network or other classification algorithm models to evaluate and predict the stability results of the label to obtain the stability results of the label.

[0077] In the embodiment of the present invention, the month-on-month difference set includes an incremental difference set and a decremental difference set, such as Figure 3 As shown in the figure, the target profile results of the target label are generated based on the month-on-month difference set, including:

[0078] Step S301: Calculate the incremental portrait result corresponding to the incremental difference set and the portrait label;

[0079] Step S302: Calculate the reduction portrait result corresponding to the reduction difference set and the portrait label;

[0080] Step S303: Obtain the historical portrait results corresponding to the historical entity set and the portrait label;

[0081] Step S304: Generate a target portrait result based on the historical portrait result, the incremental portrait result and the reduced portrait result.

[0082] In an embodiment of the present invention, the portrait analysis process is determined according to the stability result of the target tag and the stability result of the portrait tag. Specifically, if the stability results of the target tag and the portrait tag both indicate stability, a fast portrait analysis process can be used to generate the target portrait result. Among them, it can be set that when the stability value is greater than or equal to a preset value, the stability result is determined to be a stability result indicating stability, and when it is less than a preset value, the stability result is determined to be a stability result indicating instability. For the fast portrait analysis process, the target portrait result of the target tag can be generated according to the month-on-month difference set. Specifically, the incremental portrait result (P_inc) corresponding to the incremental difference set is calculated, the reduction analysis result (P_dec) of the reduction portrait is calculated, and the historical portrait result corresponding to the historical entity set is obtained, and then the difference between the incremental portrait result and the reduction portrait result is calculated, and the sum of the difference and the historical portrait result (P_hist) is used as the target portrait result (P_now), that is, P_now = P_hist + P_inc – P_dec.

[0083] In the case where at least one stability result indicating instability exists in the target tag and the portrait tag, that is, the stability result of the target tag and / or the portrait tag indicates instability, the conventional portrait analysis process is used to perform label portrait analysis, that is, the entity of the target tag is associated with the entity of the portrait tag, and the portrait data of the target tag, that is, the target portrait result, is generated. Conventional portrait analysis is based on join association operations, and is mainly calculated offline such as spark, or the Olap engine is used to perform asynchronous calculation updates during the idle time period of the machine, which has low time efficiency and limited daily data processing. However, fast portrait analysis is based on bitmap operations, and can use Olap engines such as Clickhouse (a columnar database management system for online analysis) for online real-time calculations, with a speed of sub-seconds, and due to low resource consumption (usually the amount of diff data is two orders of magnitude lower than the overall data volume), it can support almost unlimited number of portrait analyses, improving the time efficiency of portrait analysis.

[0084] In an embodiment of the present invention, the incremental difference set and the incremental portrait result corresponding to the portrait label are calculated, including: obtaining the entity set corresponding to the portrait label; associating the incremental difference set with the entity set to generate the incremental portrait result. That is, the entity set consisting of all entities corresponding to the portrait label is obtained, and then the incremental difference set and the entity set are entity joined and matched, so as to obtain the portrait result of the incremental difference set. For example, the target label is college freshman, and the age group portrait of the target label is analyzed. First, the population corresponding to the target label college freshman and the age group labels of all users are calculated, and then the age group of college freshman users can be obtained by associating the two.

[0085] Similarly, by performing join association matching on the reduced difference set and the entity set corresponding to the portrait label, the portrait result of the reduced difference set can be obtained.

[0086] In an embodiment of the present invention, after generating a target portrait result of a target tag according to the month-on-month difference set, it also includes: generating quality warning information of the target tag according to the target portrait result. That is, after generating the target portrait result of the target tag, in order to perform a quality assessment on the target tag, the target portrait result can be used as a reference indicator for instruction analysis of the target tag, and therefore, the target portrait result can be fed back to the quality warning system, such as generating quality warning information according to the target portrait result, and feeding the quality warning information back to the quality warning system to perform relevant warning analysis.

[0087] Before generating quality warning information based on the target portrait result, it is also possible to determine whether the portrait tag has a preset identifier. When it is determined that the portrait tag has a preset identifier, quality warning information is generated based on the target portrait result. The preset identifier may be an identifier indicating that the portrait tag is a core portrait, and the core portrait is used to analyze specific core attributes of the target label. For example, for a college freshman label, age group and gender can be used as core portraits. In other words, when the target portrait result is a portrait analysis result of a core portrait, quality warning information can be generated based on the target portrait result as a reference indicator for quality analysis of the target label, thereby improving the efficiency and accuracy of quality assessment.

[0088] The label portrait analysis method of the embodiment of the present invention first obtains the target entity set and the historical entity set corresponding to the target label, and generates a month-on-month difference set based on the target entity set and the historical entity set; then, by constructing the characteristic data of each label, the stability result of each label is obtained; when the stability results of the target label and the portrait label both indicate stability, the target portrait result of the target label is obtained according to the month-on-month difference set. This method realizes rapid portrait analysis of a single label through the stability result and the month-on-month difference set of the label, and can realize real-time or near real-time label portrait analysis, thereby improving timeliness; label portrait analysis based on the month-on-month difference set does not require full data association, thereby reducing resource consumption; supports the expansion of the number of portrait labels, and can flexibly expand on a large scale; and can feed back the label portrait result to the quality warning system as a quality assessment indicator of the label itself.

[0089] According to another aspect of the embodiment of the present invention, Figure 4 As shown, a label image analysis device 400 is provided, comprising:

[0090] Acquisition module 401, acquiring a target entity set and a historical entity set corresponding to a target tag;

[0091] A generation module 402 generates a month-on-month difference set of target labels according to the target entity set and the historical entity set;

[0092] Determination module 403, constructing feature data of each tag, and obtaining stability results of each tag according to the feature data, the tags including a target tag and a portrait tag corresponding to the target tag;

[0093] The analysis module 404 generates a target portrait result of the target label according to the month-on-month difference set for any portrait label when the stability result of the portrait label and the stability result of the target label both indicate stability.

[0094] In an embodiment of the present invention, generation module 402 is further used to: perform difference calculation on the target entity set and the historical entity set to obtain a month-on-month difference set, the month-on-month difference set includes an incremental difference set and / or a decremental difference set, the incremental difference set includes the target entity that is not in the historical entity set, and the decremental difference set includes the historical entity that is not in the target entity.

[0095] In an embodiment of the present invention, the generation module 402 is further used to: convert each target entity in the target entity set into a corresponding target entity code, and construct a target bitmap according to each target entity code; convert each historical entity in the historical entity set into a corresponding historical entity code, and construct a historical bitmap according to each historical entity code; calculate the difference between the target bitmap and the historical bitmap to generate an incremental bitmap of the incremental difference set and / or a decremental bitmap of the decremental difference set.

[0096] In the embodiment of the present invention, the determination module 403 is further used to: input the feature data of the label into the classification model to obtain the stability result of the label, the feature data including one or more of the label source data feature, the label calculation rule feature, the label result data feature, and the label history portrait feature;

[0097] The classification model is obtained by the following method: obtaining multiple sample labels and the stability result of each sample label; constructing feature data of each sample label; and training the classification model according to the feature data and stability result of each sample label.

[0098] In an embodiment of the present invention, the month-on-month difference set includes an incremental difference set and a decremental difference set, and the analysis module 404 is further used to: calculate the incremental portrait results corresponding to the incremental difference set and the portrait tags; calculate the decremental portrait results corresponding to the decremental difference set and the portrait tags; obtain the historical portrait results corresponding to the historical entity set and the portrait tags; and generate the target portrait results based on the historical portrait results, the incremental portrait results and the decremental portrait results.

[0099] In the embodiment of the present invention, the analysis module 404 is further used to: obtain an entity set corresponding to the portrait label; associate the incremental difference set with the entity set to generate an incremental portrait result.

[0100] In the embodiment of the present invention, the analysis module 404 is further used to: after generating a target portrait result of the target tag according to the month-on-month difference set, generate quality warning information of the target tag according to the target portrait result.

[0101] According to another aspect of an embodiment of the present invention, there is provided an electronic device, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the label portrait analysis method provided by the present invention.

[0102] According to another aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored, and when the program is executed by a processor, the label portrait analysis method provided by the present invention is implemented.

[0103] Figure 5 An exemplary system architecture 500 is shown to which the label image analysis method or label image analysis device according to an embodiment of the present invention can be applied.

[0104] like Figure 5 As shown, system architecture 500 may include terminal devices 501, 502, 503, a network 504 and a server 505. Network 504 is used to provide a medium for communication links between terminal devices 501, 502, 503 and server 505. Network 504 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0105] Users can use terminal devices 501, 502, 503 to interact with server 505 through network 504 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 501, 502, 503, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0106] The terminal devices 501 , 502 , and 503 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0107] Server 505 can be a server that provides various services, such as a background management server that provides support for shopping websites browsed by users using terminal devices 501, 502, and 503 (only as an example). The background management server can analyze and process the received data such as product information query requests, and feed back the processing results (such as target push information, product information - only as an example) to the terminal device. For example, the background management server can obtain the target entity set and historical entity set corresponding to the target tag; generate a month-on-month difference set of the target tag based on the target entity set and the historical entity set; construct feature data for each tag, and obtain the stability result of each tag based on the feature data, the tag includes a target tag and a portrait tag corresponding to the target tag; for any portrait tag, when the stability result of the portrait tag and the stability result of the target tag both indicate stability, generate a target portrait result of the target tag based on the month-on-month difference set, and feed back the target portrait result to the terminal device.

[0108] It should be noted that the label portrait analysis method provided in the embodiment of the present invention is generally executed by the server 505 , and accordingly, the label portrait analysis device is generally arranged in the server 505 .

[0109] It should be understood that Figure 5 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0110] Reference below Figure 6 , which shows a schematic diagram of the structure of a computer system 600 of a terminal device suitable for implementing an embodiment of the present invention. Figure 6 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0111] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the system 600 are also stored. The CPU 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0112] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0113] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the above-mentioned functions defined in the system of the present invention are executed.

[0114] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0115] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0116] The modules involved in the embodiments of the present invention may be implemented in software or hardware. The modules described may also be set in a processor, for example, they may be described as: a processor includes an acquisition module, a generation module, a determination module, and an analysis module. The names of these modules do not, in some cases, constitute limitations on the modules themselves, for example, the acquisition module may also be described as a "module for acquiring a target entity set and a historical entity set corresponding to a target tag."

[0117] As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device includes: obtaining a target entity set and a historical entity set corresponding to a target tag; generating a month-on-month difference set of the target tag based on the target entity set and the historical entity set; constructing feature data for each tag, and obtaining a stability result for each tag based on the feature data, the tag including a target tag and a portrait tag corresponding to the target tag; for any portrait tag, when the stability result of the portrait tag and the stability result of the target tag both indicate stability, generating a target portrait result of the target tag based on the month-on-month difference set.

[0118] According to the technical solution of the embodiment of the present invention, the label portrait analysis method of the embodiment of the present invention first obtains the target entity set and the historical entity set corresponding to the target label, and generates a month-on-month difference set based on the target entity set and the historical entity set; then, by constructing the characteristic data of each label, the stability result of each label is obtained; when the stability results of the target label and the portrait label both indicate stability, the target portrait result of the target label is obtained according to the month-on-month difference set. This method realizes rapid portrait analysis of a single label through the stability result and the month-on-month difference set of the label, and can realize real-time or near real-time label portrait analysis, thereby improving timeliness; label portrait analysis based on the month-on-month difference set does not require full data association, thereby reducing resource consumption; supports the expansion of the number of portrait labels, and can flexibly expand on a large scale; and can feed back the label portrait result to the quality warning system as a quality assessment indicator of the label itself.

[0119] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A label image analysis method, characterized in that: include: Get the target entity set and the historical entity set corresponding to the target label; Generate a month-on-month difference set of the target label according to the target entity set and the historical entity set; Constructing feature data of each tag, and obtaining a stability result of each tag according to the feature data of each tag, wherein the tags include a target tag and a portrait tag corresponding to the target tag; For any portrait tag, when the stability result of the portrait tag and the stability result of the target tag both indicate stability, a target portrait result of the target tag is generated according to the month-on-month difference set.

2. The method according to claim 1, characterized in that Generating a month-on-month difference set of the target label according to the target entity set and the historical entity set includes: The target entity set and the historical entity set are subjected to difference calculation to obtain the month-on-month difference set, wherein the month-on-month difference set includes an incremental difference set and / or a decremental difference set, wherein the incremental difference set includes target entities that are not in the historical entity set, and the decremental difference set includes historical entities that are not in the target entity.

3. The method according to claim 2, characterized in that Performing difference calculation on the target entity set and the historical entity set includes: Convert each target entity in the target entity set into a corresponding target entity code, and construct a target bitmap according to each target entity code; Convert each historical entity in the historical entity set into a corresponding historical entity code, and construct a historical bitmap according to each historical entity code; The target bitmap and the historical bitmap are subjected to difference calculation to generate an incremental bitmap of the incremental difference set and / or a decremental bitmap of the decremental difference set.

4. The method according to claim 1, characterized in that: The stability result of each tag is obtained according to the characteristic data, including: Inputting feature data of the label into a classification model to obtain a stability result of the label, wherein the feature data includes one or more of a label source data feature, a label calculation rule feature, a label result data feature, and a label history portrait feature; The classification model is obtained by the following method: Obtain multiple sample labels and stability results for each sample label; Construct feature data for each sample label; The classification model is trained based on the feature data and stability results of each sample label.

5. The method according to claim 1, characterized in that The month-on-month difference set includes an incremental difference set and a decremental difference set. Generating a target portrait result of the target tag according to the month-on-month difference set includes: Calculate the incremental portrait result corresponding to the incremental difference set and the portrait label; Calculate the reduction portrait result corresponding to the reduction difference set and the portrait label; Obtain the historical portrait results corresponding to the historical entity set and the portrait label; The target portrait result is generated according to the historical portrait result, the incremental portrait result and the reduced portrait result.

6. The method according to claim 5, characterized in that Calculating the incremental portrait result corresponding to the incremental difference set and the portrait label includes: Obtain an entity set corresponding to the portrait label; The incremental difference set is associated with the entity set to generate the incremental portrait result.

7. The method according to claim 1, characterized in that After generating the target portrait result of the target tag according to the month-on-month difference set, the method further includes: The quality warning information of the target label is generated according to the target profiling result.

8. A label image analysis device, characterized in that: include: An acquisition module is used to acquire a target entity set and a historical entity set corresponding to a target label; A generation module, generating a month-on-month difference set of the target label according to the target entity set and the historical entity set; A determination module is used to construct feature data of each tag and obtain a stability result of each tag according to the feature data of each tag, wherein the tag includes a target tag and a portrait tag corresponding to the target tag; The analysis module generates, for any portrait tag, a target portrait result of the target tag according to the month-on-month difference set when the stability result of the portrait tag and the stability result of the target tag both indicate stability.

9. An electronic device, characterized in that: include: one or more processors; a storage device 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 method according to any one of claims 1 to 7.

10. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images