A user behavior data storage and analysis method based on big data

Abstract

The present invention relates to the field of data storage technology, and in particular to a user behavior data storage and analysis method based on big data, which includes collecting behavior information of each user; obtaining a plurality of behavior tags; constructing information to be pushed for the user; screening and processing the behavior tags based on the user's behavior information for the information to be pushed, and constructing a historical tag set for each user; determining the matching coincidence and relative coincidence for a single user; determining whether to update the historical tag set based on the matching coincidence, periodically obtaining the user's behavior tags, matching the newly obtained behavior tags with the stored historical tags, and determining whether to update the historical tag set based on the matching results, monitoring user feedback for content push based on the stored user behavior information, and updating the stored data when determining that the stored behavior information cannot meet the user's needs, thereby improving the storage efficiency of the user behavior data.

Description

Technical Field

[0001] The present invention relates to the field of data storage technology, and in particular to a user behavior data storage and analysis method based on big data. Background Art

[0002] With the continuous development of computers and networks, people are increasingly relying on the Internet to obtain the content they are interested in. In order to provide users with a better user experience, operators often analyze user historical operation records to infer user interest categories in order to push relevant content to users. This requires operators to store user historical operations in the form of data in advance.

[0003] Chinese Patent Publication No.: CN103577441B, discloses a user operation data storage method and device, including: receiving a user identifier and a single user operation data corresponding to the user identifier, the single user operation data including an operation type; extracting total user operation data corresponding to the user identifier one by one from an operation database; storing the single user operation data into the total user operation data, and then storing the total user operation data into the operation database; It can be seen that the above technical solution has the following problems: each operation data of the user is stored, resulting in data redundancy, and does not consider monitoring user feedback for content push based on the stored user behavior information, and only updates the stored data when it is determined that the stored behavior information cannot meet the user's needs, which affects the storage efficiency of the user behavior data. Summary of the invention

[0004] To this end, the present invention provides a user behavior data storage and analysis method based on big data, which is used to overcome the problem that in the prior art, all operation data of users are stored, resulting in data redundancy, and no consideration is given to monitoring user feedback for content push based on stored user behavior information. The stored data is only updated when it is determined that the stored behavior information cannot meet user needs, affecting the storage efficiency of user behavior data.

[0005] To achieve the above object, the present invention provides a user behavior data storage and analysis method based on big data, comprising:

[0006] S1, collect the behavior information of each user, where the behavior information includes browsing information and purchase information;

[0007] S2, labeling the behavior information to obtain a number of behavior labels;

[0008] S3, constructing information to be pushed to the user based on the user's behavior tags;

[0009] S4, filtering the behavior tags based on the user's behavior information for the information to be pushed, storing the filtered tags to form historical tags, and constructing a historical tag set for each user;

[0010] S5, when a preset verification period is reached, determining a matching coincidence and a relative coincidence for a single user based on the matching results of the user's behavior tags within the preset verification period and each history tag in the history tag set;

[0011] S6, determining whether to update the historical tag set based on the matching overlap, including:

[0012] Determining an update method for the historical tag set based on the matching overlap deviation, wherein the update method includes adjusting a preset verification period to a corresponding value;

[0013] Or, determine whether to update the historical tag set based on the relative overlap, and mark the corresponding behavior tag as an independent tag based on the matching result between the user's behavior tag within a preset verification period and each historical tag in the historical tag set, mark the corresponding independent tag as an alternative update tag based on the frequency of occurrence of the independent tag, and determine the update method for the historical tag set based on the number of alternative update tags, wherein the update method includes storing each alternative update tag in the historical tag set to form a historical tag or arranging each alternative update tag in descending order according to the frequency of occurrence, and selecting a preset gain number of alternative update tags as historical tags in turn.

[0014] Furthermore, in S4, the process of screening the behavior tags based on the user's behavior information for the information to be pushed includes:

[0015] The number of clicks of the user on the information to be pushed is obtained based on the browsing information. If the number of clicks is greater than a preset number, the filtered tags are stored to form a history tag; if the number of clicks is less than or equal to the preset number, the behavior tags are filtered, and the behavior tags obtained based on the user's behavior information on the information to be pushed are used as verification tags, and each behavior tag is semantically compared with each verification tag to obtain semantic similarity, and the behavior tags with semantic similarity greater than the preset semantic similarity are stored to form a history tag; the behavior tags with semantic similarity less than or equal to the preset semantic similarity are determined as screening tags;

[0016] In S6, determining whether to update the historical tag set based on the matching coincidence of the user includes:

[0017] Match the behavior tags within the preset verification period with each history tag in the history tag set, calculate the ratio of the number of matching tags to the total number of behavior tags within the preset verification period, and record it as the matching coincidence;

[0018] If the matching coincidence degree is greater than the second preset matching coincidence degree, it is determined to continue to use the current historical tag set;

[0019] If the matching coincidence is less than or equal to the second preset matching coincidence and greater than the first preset matching coincidence, determining whether to update the historical tag set based on the relative coincidence;

[0020] If the matching coincidence degree is less than or equal to the first preset matching coincidence degree, an update method for the historical tag set is determined based on the matching coincidence deviation amount.

[0021] Further, determining whether to update the historical tag set based on the relative overlap includes:

[0022] Match the behavior tags within the preset verification period with the historical tags in the historical tag set, calculate the ratio of the number of matching tags to the total number of historical tags in the historical tag set, and record it as the relative overlap;

[0023] If the relative overlap is greater than the preset relative overlap, it is determined to continue using the current historical tag set, and the first preset matching overlap and the second preset matching overlap are adjusted to corresponding values ​​based on the relative overlap;

[0024] If the relative overlap is less than or equal to the preset relative overlap, the corresponding behavior tag will be marked as an independent tag based on the matching result between the user's behavior tag within the preset verification period and each historical tag in the historical tag set, and the corresponding independent tag will be marked as an alternative update tag based on the frequency of occurrence of the independent tag, and the update method for the historical tag set will be determined based on the number of alternative update tags.

[0025] Further, based on the relative overlap, the first preset matching overlap and the second preset matching overlap are adjusted to corresponding values, wherein:

[0026] The increase ranges of the first preset matching coincidence and the second preset matching coincidence determined based on the relative coincidence are proportional to the relative coincidence.

[0027] Further, an independent tag is determined, and a behavior tag within a preset verification period that matches abnormalities with each historical tag in the historical tag set is determined as an independent tag;

[0028] Determine the execution frequency of the independent tag. For a single independent tag, the number of times the single independent tag appears in the user's behavior information within a preset verification period is recorded as the single occurrence frequency;

[0029] An independent tag whose occurrence frequency is greater than a preset occurrence frequency is determined as a candidate update tag.

[0030] Furthermore, determining an update method for the historical tag set based on the number of candidate update tags includes:

[0031] If the number of candidate update tags is less than or equal to the preset update number, it is determined to store each candidate update tag into the historical tag set to form a historical tag;

[0032] If the number of candidate update tags is greater than the preset update number, the candidate update tags are arranged in descending order according to the frequency of occurrence, and a preset gain number of candidate update tags are selected as historical tags in turn, and the preset gain number is determined based on the total number of historical tags in the historical tag set.

[0033] Further, the preset gain quantity determined based on the total quantity of each historical tag in the historical tag set is inversely proportional to the total quantity of each historical tag.

[0034] Furthermore, determining an update method for the historical tag set based on the matching coincidence deviation includes:

[0035] The difference between the first preset matching coincidence degree and the matching coincidence degree is recorded as the matching coincidence deviation;

[0036] If the matching coincidence deviation is less than or equal to the preset coincidence deviation, the preset verification period is adjusted to the corresponding value;

[0037] If the matching coincidence deviation is greater than the preset coincidence deviation, an update method for the historical tag set is determined based on the screening coincidence degree.

[0038] Furthermore, based on the degree of overlap, an update method for the historical tag set is determined, including:

[0039] Matching each independent tag within a preset verification period with each screened tag, and recording the ratio of the number of the obtained matching tags to the total number of screened tags as the screening coincidence degree;

[0040] If the screening overlap degree is less than or equal to the preset screening overlap degree, determining the update method for the historical tag set based on the number of candidate update tags;

[0041] If the screening coincidence degree is greater than the preset screening coincidence degree, each matched independent tag is determined as a historical tag, and the preset semantic similarity is adjusted to a corresponding value based on the screening coincidence degree.

[0042] Further, the preset semantic similarity is adjusted to a corresponding value based on the screening overlap, wherein:

[0043] The reduction degree of the preset semantic similarity determined based on the screening overlap degree is proportional to the screening overlap degree.

[0044] Compared with the prior art, the beneficial effect of the present invention lies in periodically acquiring user behavior tags, matching the newly acquired behavior tags with the stored historical tags to determine whether to update the historical tag set based on the matching results, monitoring user feedback on content push based on stored user behavior information, and updating the stored data only when it is determined that the stored behavior information cannot meet user needs, thereby improving the storage efficiency of user behavior data.

[0045] Furthermore, the user's behavior information is periodically obtained. When a preset verification period is reached, the current behavior tag is obtained based on the user's behavior information to obtain user feedback. Whether to update the historical tag set is determined based on the matching overlap of a single user. The matching overlap represents whether the current behavior tags match the historical tags. When the matching overlap is less than or equal to the first preset matching overlap, there are a large number of newly appeared behavior tags. In this case, the currently stored behavior information cannot meet the user's needs, and it is impossible to provide the user with a push page that meets the user's preferences. The historical tag set needs to be updated, and the stored data is only updated according to the situation. In the case of real-time collection of user behavior information, the storage efficiency of user behavior data is improved.

[0046] Furthermore, when the matching overlap is less than or equal to the second preset matching overlap and greater than the first preset matching overlap, the relative overlap is used to comprehensively determine whether to update the historical tag set. The relative overlap represents the number of matched tags as a percentage of the total number of historical tags. When the relative overlap is less than or equal to the preset relative overlap, there are still many independent tags when the number of historical tags is large. In this case, the historical tag set is updated; when the relative overlap is greater than the preset relative overlap, the relative overlap is high. Due to the low number of historical tags, there are many independent tags. In this case, the judgment criteria are adjusted to further determine the judgment criteria that are suitable for the storage capacity of the current historical tag set, so as to ensure the validity of the stored historical tags and further improve the storage efficiency of user behavior data.

[0047] Furthermore, whether to update the historical tag set is determined based on the cumulative occurrence frequency of each tag in the independent tag. The independent tag is a tag in the current behavior tag that does not match the historical tag. For a single independent tag, the user's execution of the behavior with the tag within a preset verification period is recorded as the occurrence frequency of the independent tag. If the occurrence frequency is higher than the preset occurrence frequency, the independent tag is recorded as an alternative update tag to screen valid tags, count the number of alternative update tags, and determine the update method of the historical tags. When the number of alternative update tags is less than or equal to the preset update number, there are few alternative update tags and they are directly updated; when the number of alternative update tags is greater than the preset update number, there are many alternative update tags, and the alternative update tags are arranged in descending order according to their occurrence frequency, and the corresponding number of alternative update tags are selected to update the historical behavior tag group. Corresponding historical tag update methods are adopted for different situations, which improves the maintenance efficiency of user operation data and further improves the storage efficiency of user behavior data.

[0048] Furthermore, when the matching overlap is greater than the second preset matching overlap, the deviation between the current behavior label and the historical label is too large. In this case, the update method for the historical label set is determined based on the matching overlap deviation. When the matching overlap deviation is less than or equal to the preset overlap deviation, the preset verification period is too short, resulting in incomplete acquisition of the behavior label, which leads to deviations in the monitoring of user feedback. In this case, the preset verification period is adjusted to fully acquire the user's behavior information in the time domain segment. When the matching overlap deviation is greater than the preset overlap deviation, the update method for the historical label set is determined to screen out the overlap representation. The matching situation of independent tags and filtered tags is checked. When the filtering overlap is greater than the preset filtering overlap, there is an abnormality in the filtering process of behavior tags, resulting in a large number of behavior tags that should be stored being filtered out. In this case, the preset semantic similarity is adjusted to the corresponding value to redefine the filtering criteria in the subsequent screening process. When the screening overlap is less than or equal to the preset screening overlap, there is no abnormality in the filtering of behavior tags. In this case, the corresponding independent tags are added as historical tags, which improves the maintenance efficiency of user behavior data while improving the efficiency of obtaining user historical tags, and further improves the storage efficiency of user behavior data. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flowchart of the steps of the user behavior data storage and analysis method based on big data according to an embodiment of the present invention;

[0050] Figure 2 This is a logical decision diagram for determining whether to update a historical tag set based on the matching coincidence degree according to an embodiment of the present invention. DETAILED DESCRIPTION

[0051] 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.

[0052] 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.

[0053] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0054] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0055] See also Figure 1 and Figure 2 As shown, they are respectively a flowchart of a method for storing and analyzing user behavior data based on big data according to an embodiment of the present invention, and a logic determination diagram for determining whether to update a historical tag set based on matching coincidence; a method for storing and analyzing user behavior data based on big data according to an embodiment of the present invention comprises:

[0056] S1, collect the behavior information of each user, where the behavior information includes browsing information and purchase information;

[0057] S2, labeling the behavior information to obtain a number of behavior labels;

[0058] S3, constructing information to be pushed to the user based on the user's behavior tags;

[0059] S4, filtering the behavior tags based on the user's behavior information for the information to be pushed, storing the filtered tags to form historical tags, and constructing a historical tag set for each user;

[0060] S5, when a preset verification period is reached, determining a matching coincidence and a relative coincidence for a single user based on the matching results of the user's behavior tags within the preset verification period and each history tag in the history tag set;

[0061] S6, determining whether to update the historical tag set based on the matching overlap, including:

[0062] Determining an update method for the historical tag set based on the matching overlap deviation, wherein the update method includes adjusting a preset verification period to a corresponding value;

[0063] Or, determine whether to update the historical tag set based on the relative overlap, and mark the corresponding behavior tag as an independent tag based on the matching result between the user's behavior tag within a preset verification period and each historical tag in the historical tag set, mark the corresponding independent tag as an alternative update tag based on the frequency of occurrence of the independent tag, and determine the update method for the historical tag set based on the number of alternative update tags, wherein the update method includes storing each alternative update tag in the historical tag set to form a historical tag or arranging each alternative update tag in descending order according to the frequency of occurrence, and selecting a preset gain number of alternative update tags as historical tags in turn.

[0064] Specifically, there is no limitation on the specific method of labeling the behavior information to obtain a number of behavior tags. Keywords can be extracted and matched in the text information of the browsing information and the purchase information (such as the browsed web page content, product descriptions, etc.), and then these keywords are matched with pre-set tags to obtain a number of behavior tags for a single user, or the keywords extracted from the text information are directly used as behavior tags for a single user. This will not be elaborated here.

[0065] Specifically, the user's behavior tags are periodically acquired, and the newly acquired behavior tags are matched with the stored historical tags to determine whether to update the historical tag set based on the matching results. The user feedback on content push based on the stored user behavior information is monitored, and the stored data is updated only when it is determined that the stored behavior information cannot meet the user's needs, thereby improving the storage efficiency of user behavior data.

[0066] Specifically, in S4, the process of screening the behavior tags based on the user's behavior information for the information to be pushed includes:

[0067] The number of clicks of the user on the information to be pushed is obtained based on the browsing information. If the number of clicks is greater than the preset number, the filtered tags are stored and processed to form history tags; if the number of clicks is less than or equal to the preset number, the behavior tags are filtered, and the behavior tags obtained based on the user's behavior information on the information to be pushed are used as verification tags. Each behavior tag is semantically compared with each verification tag to obtain semantic similarity, and the behavior tags with semantic similarity greater than the preset semantic similarity are stored and processed to form history tags; the behavior tags with semantic similarity less than or equal to the preset semantic similarity are determined as screening tags.

[0068] Specifically, the preset number of times is selected within the interval [2, 6], and the preset semantic similarity is selected within the interval [0.2, 0.7].

[0069] Specifically, the process of semantic comparison is to perform semantic analysis on each label, perform semantic understanding on the text description of the label, perform keyword division on the semantics of each determined label to obtain a number of semantic keywords, and determine the semantic similarity as the ratio of the number of overlapping keywords in each semantic keyword of the behavior label and the verification label to the total number of semantic keywords of the corresponding behavior label and verification label.

[0070] Specifically, in S6, determining whether to update the historical tag set based on the matching coincidence of the user includes:

[0071] Match the behavior tags within the preset verification period with each history tag in the history tag set, calculate the ratio of the number of matching tags to the total number of behavior tags within the preset verification period, and record it as the matching coincidence;

[0072] If the matching coincidence degree is greater than the second preset matching coincidence degree, it is determined to continue to use the current historical tag set;

[0073] If the matching coincidence is less than or equal to the second preset matching coincidence and greater than the first preset matching coincidence, determining whether to update the historical tag set based on the relative coincidence;

[0074] If the matching coincidence degree is less than or equal to the first preset matching coincidence degree, an update method for the historical tag set is determined based on the matching coincidence deviation amount.

[0075] Specifically, the first preset matching coincidence degree P1 is selected within the interval [0.31, 0.43], and the second preset matching coincidence degree P2 is selected within the interval [0.62, 0.86].

[0076] Specifically, there is no limitation on the specific method of matching the behavior tags within the preset verification period with the historical tags in the historical tag set to determine whether they match. A semantic comparison can be performed on each tag to obtain semantic similarity, and two tags with semantic similarity greater than the preset semantic similarity are determined to be matched, and two tags with semantic similarity less than or equal to the preset semantic similarity are determined to be matched abnormally, and the behavior tags within the preset verification period with matching abnormalities are determined to be independent tags, which will not be repeated here.

[0077] Specifically, the user's behavior information is periodically obtained. When the preset verification period is reached, the current behavior tag is obtained based on the user's behavior information to obtain user feedback. The matching coincidence of a single user is used to determine whether the historical tag set is updated. The matching coincidence represents whether the current behavior tags match the historical tags. When the matching coincidence is less than or equal to the first preset matching coincidence, there are a large number of newly emerged behavior tags. In this case, the currently stored behavior information cannot meet the user's needs, and it is impossible to provide the user with a push page that meets the user's preferences. The historical tag set needs to be updated, and the stored data is only updated according to the situation. In the case of real-time collection of user behavior information, the storage efficiency of user behavior data is improved.

[0078] Specifically, determining whether to update the historical tag set based on the relative overlap includes:

[0079] Match the behavior tags within the preset verification period with the historical tags in the historical tag set, calculate the ratio of the number of matching tags to the total number of historical tags in the historical tag set, and record it as the relative overlap;

[0080] If the relative overlap is greater than the preset relative overlap, it is determined to continue using the current historical tag set, and the first preset matching overlap and the second preset matching overlap are adjusted to corresponding values ​​based on the relative overlap;

[0081] If the relative overlap is less than or equal to the preset relative overlap, the corresponding behavior tag will be marked as an independent tag based on the matching result between the user's behavior tag within the preset verification period and each historical tag in the historical tag set, and the corresponding independent tag will be marked as an alternative update tag based on the frequency of occurrence of the independent tag, and the update method for the historical tag set will be determined based on the number of alternative update tags.

[0082] Specifically, the preset relative overlap degree S0 is selected within the interval [0.29, 0.35].

[0083] Specifically, when the matching overlap is less than or equal to the second preset matching overlap and greater than the first preset matching overlap, the relative overlap is used to comprehensively determine whether to update the historical tag set. The relative overlap represents the number of matching tags as a percentage of the total number of historical tags. When the relative overlap is less than or equal to the preset relative overlap, there are still many independent tags when the number of historical tags is large. In this case, the historical tag set is updated; when the relative overlap is greater than the preset relative overlap, the relative overlap is high. Due to the low number of historical tags, there are many independent tags. In this case, the judgment criteria are adjusted to further determine the judgment criteria that are suitable for the storage capacity of the current historical tag set, so as to ensure the effectiveness of the stored historical tags and further improve the storage efficiency of user behavior data.

[0084] Specifically, based on the relative overlap, the first preset matching overlap and the second preset matching overlap are adjusted to corresponding values, wherein:

[0085] The increase ranges of the first preset matching coincidence and the second preset matching coincidence determined based on the relative coincidence are proportional to the relative coincidence.

[0086] In this embodiment, optionally,

[0087] Comparing the relative overlap with a first preset relative comparison threshold and a second preset relative comparison threshold;

[0088] If the relative overlap is less than or equal to the first preset relative comparison threshold, the first preset relative comparison threshold is adjusted to 0.92 times the initial first preset relative comparison threshold, and the second preset relative comparison threshold is adjusted to 0.92 times the initial second preset relative comparison threshold;

[0089] If the relative overlap is less than or equal to the second preset relative comparison threshold and greater than the first preset relative comparison threshold, the first preset relative comparison threshold is adjusted to 0.83 times the initial first preset relative comparison threshold, and the second preset relative comparison threshold is adjusted to 0.83 times the initial second preset relative comparison threshold;

[0090] If the relative overlap is greater than the second preset relative comparison threshold, the first preset relative comparison threshold is adjusted to 0.72 times the initial first preset relative comparison threshold, and the second preset relative comparison threshold is adjusted to 0.72 times the initial second preset relative comparison threshold;

[0091] The first preset relative comparison threshold is 1.3S0, and the second preset relative comparison threshold is 2S0.

[0092] Specifically, an independent tag is determined, and a behavior tag within a preset verification period that matches abnormalities with each historical tag in the historical tag set is determined as an independent tag;

[0093] Determine the execution frequency of the independent tag. For a single independent tag, the number of times the single independent tag appears in the user's behavior information within a preset verification period is recorded as the single occurrence frequency;

[0094] An independent tag whose occurrence frequency is greater than a preset occurrence frequency is determined as a candidate update tag.

[0095] Specifically, the preset occurrence frequency is selected within the interval [0.25N0, 0.35N0], and N0 is the total number of each behavior label of the user within the preset verification period.

[0096] Specifically, the updating method for the historical tag set is determined based on the number of candidate update tags, including:

[0097] If the number of candidate update tags is less than or equal to the preset update number, it is determined to store each candidate update tag into the historical tag set to form a historical tag;

[0098] If the number of candidate update tags is greater than the preset update number, the candidate update tags are arranged in descending order according to the frequency of occurrence, and a preset gain number of candidate update tags are selected as historical tags in turn, and the preset gain number is determined based on the total number of historical tags in the historical tag set.

[0099] Specifically, the preset update number is selected within the interval [4, 6].

[0100] Specifically, whether to update the historical tag set is determined based on the cumulative occurrence frequency of each tag in the independent tag. The independent tag is a tag in the current behavior tag that does not match the historical tag. For a single independent tag, the user's behavior with the tag within a preset verification period is recorded as the occurrence frequency of the independent tag. If the occurrence frequency is higher than the preset occurrence frequency, the independent tag is recorded as an alternative update tag to screen valid tags, count the number of alternative update tags, and determine the update method of the historical tags. When the number of alternative update tags is less than or equal to the preset update number, there are few alternative update tags and they are updated directly; when the number of alternative update tags is greater than the preset update number, there are many alternative update tags. The alternative update tags are arranged in descending order according to their occurrence frequency, and the corresponding number of alternative update tags are selected to update the historical behavior tag group. Corresponding historical tag update methods are adopted for different situations, which improves the maintenance efficiency of user operation data and further improves the storage efficiency of user behavior data.

[0101] Specifically, the preset gain quantity determined based on the total quantity of each historical tag in the historical tag set is inversely proportional to the total quantity of each historical tag.

[0102] In this embodiment, optionally,

[0103] Compare the total number of each history tag with the first preset history number and the second preset history number;

[0104] If the total number of history tags is less than or equal to the first preset history number, the preset gain number is determined to be 0.29 times the number of candidate update tags;

[0105] If the total number of history tags is less than or equal to the second preset history number and greater than the first preset history number, the preset gain number is determined to be 0.21 times the number of candidate update tags;

[0106] If the total number of history tags is greater than the second preset history number, the preset gain number is determined to be 0.11 times the number of candidate update tags;

[0107] The first preset history quantity is 18L0, and the second preset history quantity is 46L0, where L0 is the ratio of the number of newly added history tags in the previous preset verification cycle to the total number of each history tag.

[0108] Specifically, the update method for the historical tag set is determined based on the matching coincidence deviation, including:

[0109] The difference between the first preset matching coincidence degree and the matching coincidence degree is recorded as the matching coincidence deviation;

[0110] If the matching coincidence deviation is less than or equal to the preset coincidence deviation, the preset verification period is adjusted to the corresponding value;

[0111] If the matching coincidence deviation is greater than the preset coincidence deviation, an update method for the historical tag set is determined based on the screening coincidence degree.

[0112] Specifically, the preset overlap deviation is 0.68P1.

[0113] In this embodiment, optionally, the preset verification period can be adjusted to 1.19 times of the initial preset verification period.

[0114] Specifically, the update method for the historical tag set is determined based on the degree of overlap, including:

[0115] Matching each independent tag within a preset verification period with each screened tag, and recording the ratio of the number of the obtained matching tags to the total number of screened tags as the screening coincidence degree;

[0116] If the screening overlap degree is less than or equal to the preset screening overlap degree, determining the update method for the historical tag set based on the number of candidate update tags;

[0117] If the screening coincidence degree is greater than the preset screening coincidence degree, each matched independent tag is determined as a historical tag, and the preset semantic similarity is adjusted to a corresponding value based on the screening coincidence degree.

[0118] Specifically, the preset screening overlap degree H0 is selected within the interval [0.32, 0.44].

[0119] Specifically, when the matching overlap is greater than the second preset matching overlap, the deviation between the current behavior label and the historical label is too large. In this case, the update method for the historical label set is determined based on the matching overlap deviation. When the matching overlap deviation is less than or equal to the preset overlap deviation, the preset verification period is too short, resulting in incomplete acquisition of behavior labels, resulting in deviations in monitoring of user feedback. In this case, the preset verification period is adjusted to fully acquire the user's behavior information in the time domain segment. When the matching overlap deviation is greater than the preset overlap deviation, the update method for the historical label set is determined, and the overlap table is screened out. The matching situation of independent tags and filtered tags is checked. When the filtering overlap is greater than the preset filtering overlap, there is an abnormality in the filtering process of behavior tags, resulting in a large number of behavior tags that should be stored being filtered out. In this case, the preset semantic similarity is adjusted to the corresponding value to redefine the filtering standard in the subsequent screening process. When the screening overlap is less than or equal to the preset screening overlap, there is no abnormality in the filtering of behavior tags. In this case, the corresponding independent tag is added as a historical tag, which improves the efficiency of maintaining user behavior data while improving the efficiency of obtaining user historical tags, and further improves the storage efficiency of user behavior data.

[0120] Specifically, the preset semantic similarity is adjusted to a corresponding value based on the screening overlap, wherein:

[0121] The reduction degree of the preset semantic similarity determined based on the screening overlap degree is proportional to the screening overlap degree.

[0122] In this embodiment, optionally,

[0123] Comparing the screening coincidence degree with a first preset screening comparison threshold and a second preset screening comparison threshold;

[0124] If the screening coincidence degree is less than or equal to the first preset screening comparison threshold, the preset semantic similarity is adjusted to 0.95 times the current preset semantic similarity;

[0125] If the screening coincidence degree is less than or equal to the second preset screening comparison threshold and greater than the first preset screening comparison threshold, the preset semantic similarity is adjusted to 0.88 times the current preset semantic similarity;

[0126] If the screening coincidence degree is greater than the second preset screening comparison threshold, the preset semantic similarity is adjusted to 0.76 times the current preset semantic similarity;

[0127] The first preset screening comparison threshold is 1.3H0, and the second preset screening comparison threshold is 1.6H0.

[0128] 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.

[0129] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A user behavior data storage and analysis method based on big data, characterized in that: include: S1, collect the behavior information of each user, where the behavior information includes browsing information and purchase information; S2, labeling the behavior information to obtain a number of behavior labels; S3, constructing information to be pushed to the user based on the user's behavior tags; S4, filtering the behavior tags based on the user's behavior information for the information to be pushed, storing the filtered tags to form historical tags, and constructing a historical tag set for each user; S5, when a preset verification period is reached, determining a matching coincidence and a relative coincidence for a single user based on the matching results of the user's behavior tags within the preset verification period and each history tag in the history tag set; S6, determining whether to update the historical tag set based on the matching overlap, including: Determining an update method for the historical tag set based on the matching overlap deviation, wherein the update method includes adjusting a preset verification period to a corresponding value; Or, determine whether to update the historical tag set in combination with the relative overlap, and mark the corresponding behavior tag as an independent tag based on the matching result between the behavior tag of the user within a preset verification period and each historical tag in the historical tag set, mark the corresponding independent tag as an alternative update tag based on the occurrence frequency of the independent tag, and determine the update method for the historical tag set based on the number of alternative update tags, wherein the update method includes storing each alternative update tag in the historical tag set to form a historical tag or arranging each alternative update tag in descending order according to the occurrence frequency, and selecting a preset gain number of alternative update tags as historical tags in turn; In S4, the process of screening the behavior tags based on the user's behavior information for the information to be pushed includes: The number of clicks of the user on the information to be pushed is obtained based on the browsing information. If the number of clicks is greater than a preset number, the filtered tags are stored to form a history tag; if the number of clicks is less than or equal to the preset number, the behavior tags are filtered, and the behavior tags obtained based on the user's behavior information on the information to be pushed are used as verification tags, and each behavior tag is semantically compared with each verification tag to obtain semantic similarity, and the behavior tags with semantic similarity greater than the preset semantic similarity are stored to form a history tag; the behavior tags with semantic similarity less than or equal to the preset semantic similarity are determined as screening tags; In S6, determining whether to update the historical tag set based on the user's matching coincidence degree includes: Match the behavior tags within the preset verification period with each history tag in the history tag set, calculate the ratio of the number of matching tags to the total number of behavior tags within the preset verification period, and record it as the matching coincidence; If the matching coincidence degree is greater than the second preset matching coincidence degree, it is determined to continue to use the current historical tag set; If the matching coincidence is less than or equal to the second preset matching coincidence and greater than the first preset matching coincidence, determining whether to update the historical tag set based on the relative coincidence; If the matching coincidence degree is less than or equal to the first preset matching coincidence degree, an update method for the historical tag set is determined based on the matching coincidence deviation amount.

2. The user behavior data storage and analysis method based on big data according to claim 1 is characterized in that: Determining whether to update the historical tag set based on the relative overlap includes: Match the behavior tags within the preset verification period with the historical tags in the historical tag set, calculate the ratio of the number of matching tags to the total number of historical tags in the historical tag set, and record it as the relative overlap; If the relative overlap is greater than the preset relative overlap, it is determined to continue using the current historical tag set, and the first preset matching overlap and the second preset matching overlap are adjusted to corresponding values ​​based on the relative overlap; If the relative overlap is less than or equal to the preset relative overlap, the corresponding behavior tag will be marked as an independent tag based on the matching result between the user's behavior tag within the preset verification period and each historical tag in the historical tag set, and the corresponding independent tag will be marked as an alternative update tag based on the frequency of occurrence of the independent tag, and the update method for the historical tag set will be determined based on the number of alternative update tags.

3. The user behavior data storage and analysis method based on big data according to claim 2 is characterized in that: The first preset matching coincidence and the second preset matching coincidence are adjusted to corresponding values ​​based on the relative coincidence, wherein: The increase ranges of the first preset matching coincidence and the second preset matching coincidence determined based on the relative coincidence are proportional to the relative coincidence.

4. The user behavior data storage and analysis method based on big data according to claim 3 is characterized in that: Determine an independent tag, and determine the behavior tag within a preset verification period that matches abnormalities with each historical tag in the historical tag set as an independent tag; Determine the execution frequency of the independent tag. For a single independent tag, the number of times the single independent tag appears in the user's behavior information within a preset verification period is recorded as the single occurrence frequency; An independent tag whose occurrence frequency is greater than a preset occurrence frequency is determined as a candidate update tag.

5. The user behavior data storage and analysis method based on big data according to claim 4 is characterized in that: An update method for the historical tag set is determined based on the number of candidate update tags, including: If the number of candidate update tags is less than or equal to the preset update number, it is determined that each candidate update tag is stored in the historical tag set to form a historical tag; If the number of candidate update tags is greater than the preset update number, the candidate update tags are arranged in descending order according to the frequency of occurrence, and a preset gain number of candidate update tags are selected as historical tags in turn, and the preset gain number is determined based on the total number of historical tags in the historical tag set.

6. The user behavior data storage and analysis method based on big data according to claim 5 is characterized in that: The preset gain quantity determined based on the total quantity of each historical tag in the historical tag set is inversely proportional to the total quantity of each historical tag.

7. The user behavior data storage and analysis method based on big data according to claim 6 is characterized in that: The update method for the historical label set is determined based on the matching coincidence deviation, including: The difference between the first preset matching coincidence degree and the matching coincidence degree is recorded as the matching coincidence deviation; If the matching coincidence deviation is less than or equal to the preset coincidence deviation, the preset verification period is adjusted to the corresponding value; If the matching coincidence deviation is greater than the preset coincidence deviation, an update method for the historical tag set is determined based on the screening coincidence degree.

8. The user behavior data storage and analysis method based on big data according to claim 7 is characterized in that: The update method for the historical tag set is determined based on the degree of overlap, including: Matching each independent tag within a preset verification period with each screened tag, and recording the ratio of the number of the obtained matching tags to the total number of screened tags as the screening coincidence degree; If the screening overlap degree is less than or equal to the preset screening overlap degree, determining the update method for the historical tag set based on the number of candidate update tags; If the screening coincidence degree is greater than the preset screening coincidence degree, each matched independent tag is determined as a historical tag, and the preset semantic similarity is adjusted to a corresponding value based on the screening coincidence degree.

9. The method for storing and analyzing user behavior data based on big data according to claim 8 is characterized in that: The preset semantic similarity is adjusted to a corresponding value based on the screening overlap, wherein: The reduction degree of the preset semantic similarity determined based on the screening overlap degree is proportional to the screening overlap degree.

Images