Intelligent short message data management system and method based on big data

Abstract

The invention discloses a short message data intelligent management system and method based on big data, and relates to the technical field of short message data management.The management method comprises the following steps that after user authorization, the filtering process of each short message received by communication equipment is recorded, and a corresponding filtering record is generated; performing feature extraction on the set filtering rule, and performing category division on the short messages corresponding to the filtering records; based on the presentation condition of each feature in the filtering record, establishing an exception assessment rule to assess the filtering condition; capturing each short message browsing behavior of the user, summarizing each feature of short messages browsed by the user, and correcting the filtering rule; and on the basis of the corrected filtering rule, extracting corresponding characteristics of the received real-time short message, carrying out abnormity evaluation on the real-time short message, and judging whether the real-time short message is pushed to the user or not.

Description

Technical Field

[0001] The present invention relates to the technical field of SMS data management, and in particular to an intelligent SMS data management system and method based on big data. Background Art

[0002] Spam SMS usually refers to SMS messages with advertising, promotional, fraudulent and other natures sent to users without their consent or request. These spam SMS messages often involve the collection and use of personal information. Frequent receipt of spam SMS messages will cause users to feel distressed and affect their normal life. Therefore, SMS filtering is set up in SMS management to help screen and process received SMS messages, so as to identify and block unnecessary content such as spam SMS and harassing messages. Most of the existing SMS filtering rules are based on preset keywords for filtering, which can reduce the push of spam SMS to a certain extent; however, spam SMS will constantly change keywords, resulting in the preset keywords not being able to completely cover all spam SMS, or some normal SMS may also contain preset keywords, resulting in normal messages being unable to be received; as the frequency of false detection increases, it will also affect users' information acquisition and usage experience. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent management system and method for SMS data based on big data to solve the problems raised in the prior art.

[0004] To achieve the above object, the present invention provides the following technical solution: an intelligent management method for SMS data based on big data, the management method comprising the following steps: Step S100: after user authorization, the filtering process of each SMS message received by the communication device is recorded to generate a corresponding filtering record; based on the filtering results presented by each filtering record, feature extraction is performed on the set filtering rules; Step S200: Analyze the feature inclusion of any filtering record, and classify the short messages corresponding to the filtering record into categories; and establish an abnormal evaluation rule based on the presentation of each feature in the filtering record to evaluate the filtering situation; Step S300: Capture each SMS browsing behavior of the user, analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; summarize the features of the SMS browsed by the user, and modify the set filtering rules; Step S400: Based on the modified filtering rules, extract corresponding features from the received real-time SMS, and perform abnormality assessment on the real-time SMS; and determine whether to push the real-time SMS to the user based on the abnormality assessment result.

[0005] Furthermore, step S100 includes the following steps: Step S101: Whenever a user's communication device receives a text message, the text message content of the received text message is read, and a preset text message filtering rule is called to judge the text message content to obtain a filtering result of the received text message, and a filtering record is generated to store the text message content and filtering result of the received text message; if the received text message is not pushed to the user after the text message filtering, the filtering record of the received text message is marked as abnormal; Step S102: randomly selecting a filtering record, and if there is an abnormal mark in the filtering record, performing word segmentation processing on the SMS content in the filtering record to generate a number of keywords; Step S103: Acquire all keywords of each filtering record with an abnormal mark, extract any two keywords for similarity comparison, and if the obtained similarity exceeds a set similarity threshold, set the two keywords as the same keywords, and generate a corresponding keyword set for each type of the same keywords; Step S104: arbitrarily select a keyword set, obtain the number of keywords contained in the keyword set, and set the number of keywords in the i-th keyword set as M. i , calculate the proportion of the number of keyword sets in the i-th category α i =M i / A, where A is the number of filtered records with abnormal marks; Step S105: Set a quantity ratio threshold as α max , randomly select a filter record with an abnormal mark, obtain the keyword set where the jth keyword in the filter record with the abnormal mark is located, set the jth keyword to correspond to the i-th keyword set, if α i >α max , then the j-th keyword is set as a feature of the SMS filtering rule. If the quantity ratio of the keyword set corresponding to any keyword is less than the quantity ratio threshold, then the keyword with the largest quantity ratio is selected as a feature of the SMS filtering rule; the features set in all filtering records with abnormal marks are summarized to obtain the feature set of the SMS filtering rule; by identifying the adjustment of the SMS filtering rule, it is beneficial to the calculation of subsequent abnormal evaluation and the development of feature adjustment.

[0006] Further, step S200 includes the following steps: Step S201: randomly select a filtering record, extract keywords from the SMS content stored in the filtering record, and compare the similarity between each keyword and any feature in the feature set. If the obtained similarity exceeds the set similarity threshold, match the compared keyword with the keyword set corresponding to the compared feature, and obtain the quantity ratio of the corresponding keyword set. Step S202: Obtain the proportion of the number of keyword sets matched by each keyword, sort the keyword sets from large to small according to the proportion of the number, select the keyword set with the largest proportion of the number as the first feature type of the filtering record, and so on, take the keyword set with the position b as the bth feature type of the filtering record; the feature types of different levels represent the influence of the corresponding keyword sets, which are used as the criterion for determining whether it is a spam message first. If the number of keywords involved in the message belonging to the first feature type is larger, the probability of being determined as a spam message is higher; Step S203: arbitrarily select a keyword set with position b and the number of keywords corresponding to the filtering record is N. b , set the proportion of the keyword set with position b to α b , build an evaluation model: ; Wherein, b1 is a positive integer and b1∈(1,e), e is the number of keyword sets to which the filtering record belongs, and N is the total number of keywords extracted from the SMS content of the filtering record; the evaluation value P of the filtering record is calculated; the proportion of quantity and the number of keywords appearing are the most direct criteria for determining whether it is a spam SMS, and based on the proportion of relevant keywords in the SMS content, a more accurate evaluation value can be obtained, helping users to accurately identify whether it is a spam SMS; Step S204: If there is an abnormal mark in the filtering record, the evaluation value P of the filtering record is set as the abnormal evaluation value; the abnormal evaluation values ​​of all filtering records with abnormal marks are obtained, and the abnormal evaluation value P with the smallest value is selected. min As an exception assessment rule to determine whether a text message is filtered.

[0007] Further, step S300 includes the following steps: Step S301: Whenever a user browses a text message, obtain the evaluation value P of the filtering record corresponding to the text message, and set the abnormal evaluation value for determining whether the text message is filtered to P min , if P>P min , extract the SMS content in the SMS and segment the SMS content to obtain several keywords; Step S302: Compare each keyword with the feature set of the filtering rule to obtain several features contained in the text message, arbitrarily select the kth feature, obtain the keyword set corresponding to the jth feature, and obtain the number of keyword sets corresponding to the kth feature as α k , get the number of filtered records with abnormal marks as N, according to the formula: ; Calculate the proportion of the corrected number of keyword sets corresponding to the kth feature α ’ k ; Set the quantity ratio threshold to α max , if α ’ k <α max , then the kth feature is removed from the feature set of the filtering rule; if it is identified that the user browses the content of the SMS identified as spam, it means that there is an error in the current filtering rule or the user has a wrong click behavior, then it is necessary to adjust the number of corresponding keywords and re-perform feature recognition, which can help reduce the probability of identifying normal SMS as spam; Step S303: when there is a text message pushed to the user, and the user does not browse the pushed text message, each keyword of the pushed text message is extracted, and a similarity comparison is performed between any keyword and each feature. If the obtained similarity is less than a set similarity threshold, the keyword is set as the expected feature; and a feature set of the pushed text message and an expected feature set are obtained; Step S304: If the feature set of the pushed SMS is empty, then select any desired feature from the desired feature set, compare the desired feature with the keywords in any filtering record for similarity, and obtain the number of identical keywords corresponding to the desired feature as W. sim , according to the formula: ; Where N is the number of filtered records with abnormal marks; the proportion of the number of corrections corresponding to the same keyword of the expected feature is calculated as α ’ ; If α ’ >α max , then the expected feature is set as the new feature of the filtering rule; the user does not browse the pushed SMS, and also needs to be adjusted to help the user reduce the probability of misjudgment due to new keywords in spam SMS, and adaptively improve the accuracy of recognition; Step S305: extract features from each SMS message browsed by the user and pushed by the communication device, obtain a new feature set and overwrite the feature set of the filtering rule, and modify the original filtering rule.

[0008] Furthermore, step S400 includes the following steps: Step S401: when a user's communication device receives a real-time text message in real time, the text message content of the real-time text message is obtained to obtain a number of keywords, and each keyword is compared with a feature set of the modified filtering rule to generate a number of features contained in the real-time text message; Step S402: Obtain the quantity ratio of the keyword set corresponding to each feature, and sort the features from large to small according to the quantity ratio; call the evaluation model to evaluate the real-time SMS to obtain an evaluation value of P new ; Get the abnormal evaluation value P to determine whether the text message is filtered min , if P new <P min , then push the real-time SMS to the user, if P new >P min , the real-time short message is filtered.

[0009] In order to better implement the above method, an intelligent SMS data management system is also proposed, which includes a historical SMS analysis module, an SMS classification evaluation module, a user behavior analysis module and an SMS real-time analysis module; The historical SMS analysis module is used to record the filtering process of each SMS received by the communication device after user authorization and generate corresponding filtering records; based on the filtering results presented by each filtering record, feature extraction is performed on the set filtering rules; The SMS classification and evaluation module is used to analyze the feature inclusion of any filtering record and classify the SMS corresponding to the filtering record into categories; based on the presentation of each feature in the filtering record, an abnormal evaluation rule is established to evaluate the filtering situation; The SMS filtering adjustment module is used to capture each SMS browsing behavior of the user, analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; summarize the features of the SMS browsed by the user, and modify the set filtering rules; The SMS real-time filtering module is used to extract corresponding features of the received real-time SMS based on the revised filtering rules, and perform abnormality evaluation on the real-time SMS; based on the abnormality evaluation result, it is determined whether to push the real-time SMS to the user.

[0010] Further, the historical SMS analysis module includes a historical filtering collection unit and a filtering rule extraction unit; The historical filtering collection unit is used to record the filtering process of each SMS received by the communication device after user authorization and generate corresponding filtering records; the filtering rule extraction unit is used to extract features of the set filtering rules based on the filtering results presented by each filtering record.

[0011] Further, the SMS classification and evaluation module includes an SMS type classification unit and an SMS anomaly evaluation unit; The SMS type classification unit is used to analyze the feature inclusion of any filtering record and classify the SMS corresponding to the filtering record; the SMS anomaly evaluation unit is used to establish anomaly evaluation rules to evaluate the filtering situation based on the presentation of each feature in the filtering record.

[0012] Further, the SMS filtering adjustment module includes a user behavior analysis unit and an identification rule correction unit; The user behavior analysis unit is used to capture each SMS browsing behavior of the user and analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; the identification rule correction unit is used to summarize the various features of the SMS browsed by the user and correct the set filtering rules.

[0013] Furthermore, the SMS real-time filtering module includes a real-time SMS analysis unit and an SMS identification and push unit; The real-time SMS analysis unit is used to extract corresponding features of the received real-time SMS based on the revised filtering rules and perform abnormality evaluation on the real-time SMS; the SMS identification and push unit is used to determine whether to push the real-time SMS to the user based on the abnormality evaluation result.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention can adaptively modify the keywords for judging junk text messages, and can call the optimal filtering rules to identify junk text messages at any time, thereby reducing the occurrence of false detections due to simple keyword judgments, optimizing the filtering of junk text messages, and improving the user's text message acquisition experience; 2. The present invention converts simple keyword recognition into abnormal evaluation value calculation, and assigns different evaluation proportions according to the historical recognition of different keywords, which can avoid the situation where normal text messages are directly filtered because of the presence of keywords, and effectively improve the filtering accuracy of junk text messages; 3. The present invention improves the simple keyword comparison in conventional technical means, improves the accuracy of identifying and filtering junk text messages, better helps users filter effective information, and improves users' information acquisition experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of the steps of an intelligent management method for SMS data based on big data; Figure 2 The figure is a structural diagram of an intelligent management system for SMS data based on big data. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Example: Figure 1 to Figure 2 As shown, the present invention provides an intelligent management method for SMS data based on big data, and the management method comprises the following steps: Step S100: after user authorization, the filtering process of each SMS message received by the communication device is recorded to generate a corresponding filtering record; based on the filtering results presented by each filtering record, feature extraction is performed on the set filtering rules; Wherein, step S100 includes the following steps: Step S101: Whenever a user's communication device receives a text message, the text message content of the received text message is read, and a preset text message filtering rule is called to judge the text message content to obtain a filtering result of the received text message, and a filtering record is generated to store the text message content and filtering result of the received text message; if the received text message is not pushed to the user after the text message filtering, the filtering record of the received text message is marked as abnormal; Step S102: randomly selecting a filtering record, and if there is an abnormal mark in the filtering record, performing word segmentation processing on the SMS content in the filtering record to generate a number of keywords; Step S103: Acquire all keywords of each filtering record with an abnormal mark, extract any two keywords for similarity comparison, and if the obtained similarity exceeds a set similarity threshold, set the two keywords as the same keywords, and generate a corresponding keyword set for each type of the same keywords; Step S104: arbitrarily select a keyword set, obtain the number of keywords contained in the keyword set, and set the number of keywords in the i-th keyword set as M. i , calculate the proportion of the number of keyword sets in the i-th category α i =M i / A, where A is the number of filtered records with abnormal marks; Step S105: Set a quantity ratio threshold as α max , randomly select a filter record with an abnormal mark, obtain the keyword set where the jth keyword in the filter record with the abnormal mark is located, set the jth keyword to correspond to the i-th keyword set, if α i >α max , then the j-th keyword is set as a feature of the SMS filtering rule. If the quantity ratio of the keyword set corresponding to any keyword is less than the quantity ratio threshold, then the keyword with the largest quantity ratio is selected as a feature of the SMS filtering rule; the features set in all the filtering records with abnormal marks are summarized to obtain the feature set of the SMS filtering rule.

[0018] Step S200: Analyze the feature inclusion of any filtering record, and classify the short messages corresponding to the filtering record into categories; and establish an abnormal evaluation rule based on the presentation of each feature in the filtering record to evaluate the filtering situation; Wherein, step S200 includes the following steps: Step S201: randomly select a filtering record, extract keywords from the SMS content stored in the filtering record, and compare the similarity between each keyword and any feature in the feature set. If the obtained similarity exceeds the set similarity threshold, match the compared keyword with the keyword set corresponding to the compared feature, and obtain the quantity ratio of the corresponding keyword set. Step S202: Obtain the proportion of the number of keyword sets matched by each keyword, sort the keyword sets from large to small according to the proportion of the number, select the keyword set with the largest proportion of the number as the first feature type of the filtering record, and so on, select the keyword set with the position b as the bth feature type of the filtering record; Step S203: arbitrarily select a keyword set with position b and the number of keywords corresponding to the filtering record is N. b , set the proportion of the keyword set with position b to α b , build an evaluation model: ; Wherein, b1 is a positive integer and b1∈(1,e), e is the number of keyword sets to which the filtering record belongs, and N is the total number of keywords extracted from the SMS content of the filtering record; the evaluation value P of the filtering record is calculated; Example 1: 10 keywords are extracted from a filtering record, of which 5 keywords are matched with 3 types of keyword sets, 2 of which correspond to the first type of keyword set, 2 correspond to the second type of keyword set, and 1 corresponds to the third type of keyword set; the proportions of the three types of keyword sets are set to 90%, 70% and 50% respectively, and the evaluation value P is calculated to be 5 / 10×(2 3 ×90%+2 2 ×70%+1 1 ×50%)=0.5×10.5=5.25; the abnormal evaluation value is set to 5, 5.25>5, so the text message corresponding to the filtering record will be directly filtered; Step S204: If there is an abnormal mark in the filtering record, the evaluation value P of the filtering record is set as the abnormal evaluation value; the abnormal evaluation values ​​of all filtering records with abnormal marks are obtained, and the abnormal evaluation value P with the smallest value is selected. min As an exception assessment rule to determine whether a text message is filtered.

[0019] Step S300: Capture each SMS browsing behavior of the user, analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; summarize the features of the SMS browsed by the user, and modify the set filtering rules; Wherein, step S300 includes the following steps: Step S301: Whenever a user browses a text message, obtain the evaluation value P of the filtering record corresponding to the text message, and set the abnormal evaluation value for determining whether the text message is filtered to P min , if P>P min , extract the SMS content in the SMS and segment the SMS content to obtain several keywords; Step S302: Compare each keyword with the feature set of the filtering rule to obtain several features contained in the text message, arbitrarily select the kth feature, obtain the keyword set corresponding to the jth feature, and obtain the number of keyword sets corresponding to the kth feature as α k , get the number of filtered records with abnormal marks as N, according to the formula: ; Calculate the proportion of the corrected number of keyword sets corresponding to the kth feature α ’ k ; Set the quantity ratio threshold to α max , if α ’ k <α max , then the k-th feature is removed from the feature set of the filtering rule; Example 2: It is assumed that a normal filtering record contains one feature, the number of keyword sets corresponding to the feature accounts for 60%, and the number of filtering records with abnormal marks is set to 10, so the correction number accounts for (60%×10-1) / (10-1)=55.55%. If the number account threshold is 56%, because 55.55%<56%, the included feature can be removed from the filtering rule; Step S303: when there is a text message pushed to the user, and the user does not browse the pushed text message, each keyword of the pushed text message is extracted, and a similarity comparison is performed between any keyword and each feature. If the obtained similarity is less than a set similarity threshold, the keyword is set as the expected feature; and a feature set of the pushed text message and an expected feature set are obtained; Step S304: If the feature set of the pushed SMS is empty, then select any desired feature from the desired feature set, compare the desired feature with the keywords in any filtering record for similarity, and obtain the number of identical keywords corresponding to the desired feature as W. sim , according to the formula: ; Where N is the number of filtered records with abnormal marks; the proportion of the number of corrections corresponding to the same keyword of the expected feature is calculated as α ’ ; If α ’ >α max , then the expected feature is set as a new feature of the filtering rule; Step S305: extract features from each SMS message browsed by the user and pushed by the communication device, obtain a new feature set and overwrite the feature set of the filtering rule, and modify the original filtering rule.

[0020] Step S400: extracting corresponding features of the received real-time SMS based on the revised filtering rules, and performing an abnormality assessment on the real-time SMS; and determining whether to push the real-time SMS to the user based on the abnormality assessment result; Wherein, step S400 includes the following steps: Step S401: when a user's communication device receives a real-time text message in real time, the text message content of the real-time text message is obtained to obtain a number of keywords, and each keyword is compared with a feature set of the modified filtering rule to generate a number of features contained in the real-time text message; Step S402: Obtain the quantity ratio of the keyword set corresponding to each feature, and sort the features from large to small according to the quantity ratio; call the evaluation model to evaluate the real-time SMS to obtain an evaluation value of P new ; Get the abnormal evaluation value P to determine whether the text message is filteredmin , if P new <P min , then push the real-time SMS to the user, if P new >P min , the real-time short message is filtered.

[0021] An intelligent SMS data management system, the management system includes a historical SMS analysis module, an SMS classification evaluation module, a user behavior analysis module and an SMS real-time analysis module; The historical SMS analysis module is used to record the filtering process of each SMS received by the communication device after user authorization and generate corresponding filtering records; based on the filtering results presented by each filtering record, feature extraction is performed on the set filtering rules; The SMS classification and evaluation module is used to analyze the feature inclusion of any filtering record and classify the SMS corresponding to the filtering record into categories; based on the presentation of each feature in the filtering record, an abnormal evaluation rule is established to evaluate the filtering situation; The SMS filtering adjustment module is used to capture each SMS browsing behavior of the user, analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; summarize the features of the SMS browsed by the user, and modify the set filtering rules; The SMS real-time filtering module is used to extract corresponding features of the received real-time SMS based on the revised filtering rules, and perform abnormality evaluation on the real-time SMS; based on the abnormality evaluation result, it is determined whether to push the real-time SMS to the user.

[0022] Among them, the historical SMS analysis module includes a historical filtering collection unit and a filtering rule extraction unit; The historical filtering collection unit is used to record the filtering process of each SMS received by the communication device after user authorization and generate corresponding filtering records; the filtering rule extraction unit is used to extract features of the set filtering rules based on the filtering results presented by each filtering record.

[0023] Among them, the SMS classification and evaluation module includes an SMS type classification unit and an SMS anomaly evaluation unit; The SMS type classification unit is used to analyze the feature inclusion of any filtering record and classify the SMS corresponding to the filtering record; the SMS anomaly evaluation unit is used to establish anomaly evaluation rules to evaluate the filtering situation based on the presentation of each feature in the filtering record.

[0024] Among them, the SMS filtering adjustment module includes a user behavior analysis unit and an identification rule correction unit; The user behavior analysis unit is used to capture each SMS browsing behavior of the user and analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; the identification rule correction unit is used to summarize the various features of the SMS browsed by the user and correct the set filtering rules.

[0025] Among them, the SMS real-time filtering module includes a real-time SMS analysis unit and an SMS identification and push unit; The real-time SMS analysis unit is used to extract corresponding features of the received real-time SMS based on the revised filtering rules and perform abnormality evaluation on the real-time SMS; the SMS identification and push unit is used to determine whether to push the real-time SMS to the user based on the abnormality evaluation result.

[0026] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for intelligent management of SMS data based on big data, characterized in that: The management method comprises the following steps: Step S100: after user authorization, the filtering process of each SMS message received by the communication device is recorded to generate a corresponding filtering record; based on the filtering results presented by each filtering record, feature extraction is performed on the set filtering rules; Step S200: Analyze the feature inclusion of any filtering record, and classify the short messages corresponding to the filtering record into categories; and establish an abnormal evaluation rule to evaluate the filtering situation based on the presentation of each feature in the filtering record; Step S300: Capture each SMS browsing behavior of the user, analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; summarize the features of the SMS browsed by the user, and modify the set filtering rules; Step S400: Based on the modified filtering rules, extract corresponding features from the received real-time SMS, and perform abnormality assessment on the real-time SMS; and determine whether to push the real-time SMS to the user based on the abnormality assessment result.

2. The method for intelligent management of SMS data based on big data according to claim 1, characterized in that: The step S100 includes the following steps: Step S101: Whenever a user's communication device receives a text message, the text message content of the received text message is read, and a preset text message filtering rule is called to judge the text message content to obtain a filtering result of the received text message, and a filtering record is generated to store the text message content and filtering result of the received text message; if the received text message is not pushed to the user after the text message filtering, the filtering record of the received text message is marked as abnormal; Step S102: randomly selecting a filtering record, and if there is an abnormal mark in the filtering record, performing word segmentation processing on the SMS content in the filtering record to generate a number of keywords; Step S103: Acquire all keywords of each filtering record with an abnormal mark, extract any two keywords for similarity comparison, and if the obtained similarity exceeds a set similarity threshold, set the two keywords as the same keywords, and generate a corresponding keyword set for each type of the same keywords; Step S104: arbitrarily select a keyword set, obtain the number of keywords contained in the keyword set, and set the number of keywords in the i-th keyword set as M. i , calculate the proportion of the number of keyword sets in the i-th category α i =M i / A, where A is the number of filtered records with abnormal marks; Step S105: Set a quantity ratio threshold as α max , randomly select a filter record with an abnormal mark, obtain the keyword set where the jth keyword in the filter record with the abnormal mark is located, set the jth keyword to correspond to the i-th keyword set, if α i >α max , then the j-th keyword is set as a feature of the SMS filtering rule. If the quantity ratio of the keyword set corresponding to any keyword is less than the quantity ratio threshold, then the keyword with the largest quantity ratio is selected as a feature of the SMS filtering rule; the features set in all the filtering records with abnormal marks are summarized to obtain the feature set of the SMS filtering rule.

3. The method for intelligent management of SMS data based on big data according to claim 2, characterized in that: The step S200 includes the following steps: Step S201: randomly select a filtering record, extract keywords from the SMS content stored in the filtering record, and compare the similarity between each keyword and any feature in the feature set. If the obtained similarity exceeds the set similarity threshold, match the compared keyword with the keyword set corresponding to the compared feature, and obtain the quantity ratio of the corresponding keyword set. Step S202: Obtain the proportion of the number of keyword sets matched by each keyword, sort the keyword sets from large to small according to the proportion of the number, select the keyword set with the largest proportion of the number as the first feature type of the filtering record, and so on, select the keyword set with the position b as the bth feature type of the filtering record; Step S203: arbitrarily select a keyword set with position b and the number of keywords corresponding to the filtering record is N. b , set the proportion of the keyword set with position b to α b , build an evaluation model: ； Wherein, b1 is a positive integer and b1∈(1,e), e is the number of keyword sets to which the filtering record belongs, and N is the total number of keywords extracted from the SMS content of the filtering record; the evaluation value P of the filtering record is calculated; Step S204: If there is an abnormal mark in the filtering record, the evaluation value P of the filtering record is set as the abnormal evaluation value; the abnormal evaluation values ​​of all filtering records with abnormal marks are obtained, and the abnormal evaluation value P with the smallest value is selected. min As an exception assessment rule to determine whether a text message is filtered.

4. The method for intelligent management of SMS data based on big data according to claim 3, characterized in that: The step S300 includes the following steps: Step S301: Whenever a user browses a text message, obtain the evaluation value P of the filtering record corresponding to the text message, and set the abnormal evaluation value for determining whether the text message is filtered to P min , if P>P min , extract the SMS content in the SMS and segment the SMS content to obtain several keywords; Step S302: Compare each keyword with the feature set of the filtering rule to obtain several features contained in the text message, arbitrarily select the kth feature, obtain the keyword set corresponding to the jth feature, and obtain the number of keyword sets corresponding to the kth feature as α k , get the number of filtered records with abnormal marks as N, according to the formula: ； Calculate the proportion of the corrected number of keyword sets corresponding to the kth feature α ’ k ; Set the quantity ratio threshold to α max , if α ’ k <α max , then the k-th feature is removed from the feature set of the filtering rule; Step S303: when there is a text message pushed to the user, and the user does not browse the pushed text message, each keyword of the pushed text message is extracted, and a similarity comparison is performed between any keyword and each feature. If the obtained similarity is less than a set similarity threshold, the keyword is set as the expected feature; and a feature set of the pushed text message and an expected feature set are obtained; Step S304: If the feature set of the pushed SMS is empty, then select any desired feature from the desired feature set, compare the desired feature with the keywords in any filtering record for similarity, and obtain the number of identical keywords corresponding to the desired feature as W. sim , according to the formula: ； Where N is the number of filtered records with abnormal marks; the proportion of the number of corrections corresponding to the same keyword of the expected feature is calculated as α ’ ; If α ’ >α max , then the expected feature is set as a new feature of the filtering rule; Step S305: extract features from each SMS message browsed by the user and pushed by the communication device, obtain a new feature set and overwrite the feature set of the filtering rule, and modify the original filtering rule.

5. The method for intelligent management of SMS data based on big data according to claim 4, characterized in that: The step S400 includes the following steps: Step S401: when a user's communication device receives a real-time text message in real time, the text message content of the real-time text message is obtained to obtain a number of keywords, and each keyword is compared with a feature set of the modified filtering rule to generate a number of features contained in the real-time text message; Step S402: Obtain the quantity ratio of the keyword set corresponding to each feature, and sort the features from large to small according to the quantity ratio; call the evaluation model to evaluate the real-time SMS to obtain an evaluation value of P new ; Get the abnormal evaluation value P to determine whether the text message is filtered min , if P new <P min , then push the real-time SMS to the user, if P new >P min , the real-time short message is filtered.

6. An intelligent SMS data management system, used to execute an intelligent SMS data management method based on big data as claimed in any one of claims 1 to 5, characterized in that: The management system includes a historical SMS analysis module, a SMS classification evaluation module, a user behavior analysis module and a SMS real-time analysis module; The historical SMS analysis module is used to record the filtering process of each SMS received by the communication device after user authorization, and generate corresponding filtering records; based on the filtering results presented by each filtering record, feature extraction is performed on the set filtering rules; The SMS classification evaluation module is used to analyze the feature inclusion of any filtering record and classify the SMS corresponding to the filtering record into categories; based on the presentation of each feature in the filtering record, an abnormal evaluation rule is established to evaluate the filtering situation; The SMS filtering adjustment module is used to capture each SMS browsing behavior of the user, analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; summarize the features of the SMS browsed by the user, and modify the set filtering rules; The SMS real-time filtering module is used to extract corresponding features of the received real-time SMS based on the revised filtering rules and perform abnormality assessment on the real-time SMS; Based on the abnormality assessment result, it is determined whether to push the real-time SMS to the user.

7. The intelligent management system for short message data according to claim 6, characterized in that: The historical SMS analysis module includes a historical filtering collection unit and a filtering rule extraction unit; The historical filtering collection unit is used to record the filtering process of each text message received by the communication device after user authorization and generate corresponding filtering records; the filtering rule extraction unit is used to extract features of the set filtering rules based on the filtering results presented by each filtering record.

8. The intelligent management system for short message data according to claim 6, characterized in that: The SMS classification and evaluation module includes an SMS type classification unit and an SMS anomaly evaluation unit; The SMS type classification unit is used to analyze the feature inclusion of any filtering record and classify the SMS corresponding to the filtering record; the SMS anomaly evaluation unit is used to establish anomaly evaluation rules to evaluate the filtering situation based on the presentation of each feature in the filtering record.

9. The intelligent management system for short message data according to claim 6, characterized in that: The SMS filtering adjustment module includes a user behavior analysis unit and an identification rule correction unit; The user behavior analysis unit is used to capture each SMS browsing behavior of the user and analyze the correlation between the features contained in the SMS browsed by the user and the filtering rules; the identification rule correction unit is used to summarize the various features of the SMS browsed by the user and correct the set filtering rules.

10. The intelligent management system for short message data according to claim 6, characterized in that: The SMS real-time filtering module includes a real-time SMS analysis unit and an SMS identification and push unit; The real-time SMS analysis unit is used to extract corresponding features of the received real-time SMS based on the revised filtering rules and perform abnormality assessment on the real-time SMS; The SMS identification and pushing unit is used to determine whether to push the real-time SMS to the user based on the abnormality assessment result.

Images