Communication text information conversion method and system

Abstract

The invention discloses a communication text information conversion method and system, and relates to the technical field of short message recognition, and the method comprises the steps: carrying out the component recognition of short message information, and obtaining information component data; analyzing the information component data, and determining the category of the short message; carrying out key information reading on the short message information to obtain a credibility value; judging text data and link data in the information component data, and determining the danger of the text data and the danger of the link data; based on the credibility value of the short message information, the dangerousness of the text data and the dangerousness of the link data, determining a dangerousness index of the short message information; determining a cheating index of the user on the corresponding short message category based on the short message category; and comparing the risk index of the short message information with the cheating index on the corresponding short message category, marking the short message information of which the risk index is greater than or equal to the cheating index on the corresponding short message category as a dangerous short message, and intercepting the dangerous short message. The method has the effect of improving the identification accuracy of the harassment short message.

Description

Technical Field

[0001] The present application relates to the technical field of short message recognition, and in particular to a method and system for converting communication text information. Background Art

[0002] Communication text information conversion is the process of converting text data into another form or format. With the rapid development of mobile communication technology, mobile phones have become an indispensable communication tool in people's daily lives. However, with the popularization of mobile phone use, the problem of harassing text messages has become increasingly serious. Therefore, how to improve the accuracy of intercepting harassing text messages by converting the text information in communication needs to be solved urgently.

[0003] In the related art, machine learning algorithms are used to analyze and classify text message contents to identify whether text messages of the same type are spam messages. However, criminals may add interference symbols to text messages, use traditional Chinese characters, homophonic characters, split characters and other replacement methods to change the text message content, thereby avoiding detection by classification models, making the identification of spam text messages inaccurate and reducing the user experience. Therefore, a communication text information conversion method and system are urgently needed to improve the recognition rate and interception effect of spam text messages. Summary of the invention

[0004] In order to improve the recognition accuracy of harassing text messages, the present application provides a communication text information conversion method and system.

[0005] In a first aspect, the present application provides a communication text information conversion method, which adopts the following technical solution:

[0006] A communication text information conversion method, comprising:

[0007] Acquire the SMS message received by the user, and identify the components of the SMS message to obtain information component data; the information component data includes text data and link data;

[0008] Determine whether the information component data contains text data, and if the information component data contains text data, perform text information conversion analysis on the text data to determine the SMS category;

[0009] If the information component data does not contain text data, the SMS category of the SMS message is determined to be a pure link category;

[0010] Read the key information of the SMS message and match the key information with the built-in credibility table to obtain the credibility value;

[0011] Judging the text data in the information component data to determine the danger of the text data;

[0012] Judging the linked data in the information component data to determine the danger of the linked data;

[0013] Determine the risk index of the SMS message based on the credibility value of the SMS message, the risk of the text data, and the risk of the link data;

[0014] Based on the SMS category of the SMS information, the user's fraud index is matched to obtain the user's fraud index in the corresponding SMS category;

[0015] Compare the danger index of the SMS message with the fraud index of the corresponding SMS category, mark the SMS message with a danger index less than the fraud index of the corresponding SMS category as a safe SMS and present it, and mark the SMS message with a danger index greater than or equal to the fraud index of the corresponding SMS category as a dangerous SMS and intercept it.

[0016] Preferably, semantic understanding is performed on the text data to obtain first semantic data, and semantic integrity judgment is performed on the first semantic data;

[0017] If the first semantic data is complete, then performing a core content judgment on the text data to determine whether the core content is product information;

[0018] If the core content is product information, the SMS category is determined to be advertising and promotion;

[0019] If the central content is not product information, the text data is matched with sensitive words to determine whether sensitive words exist in the text data;

[0020] If there are sensitive words in the text data, the SMS is classified as spam;

[0021] If there are no sensitive words in the text data, performing an inductive judgment on the text data according to the first semantic data;

[0022] If the text data is misleading to the user's behavior, the SMS category is determined to be misleading;

[0023] If the first semantic data is not complete, obtaining syllable data of the text data, generating corresponding voice data according to the syllable data, and converting the voice data into text to obtain recognized text data;

[0024] Performing semantic understanding on the recognized text data to obtain second semantic data;

[0025] A semantic integrity judgment is performed on the second semantic data to determine the integrity of the second semantic data, and the SMS category of the recognized text data is determined based on the integrity of the second semantic data.

[0026] Preferably, the second semantic data is subjected to semantic integrity judgment, and if the second semantic data is complete, the SMS category of the recognized text data is determined;

[0027] If the second semantic data is not complete, the second semantic data is divided according to the completeness of the sentence to obtain second semantic sub-data with a sentence order; the second semantic sub-data includes second complete sub-data with complete semantics and second incomplete sub-data without a complete sentence;

[0028] The recognized character data is split according to the second semantic sub-data to obtain recognized character sub-data corresponding to the second semantic sub-data; the recognized character sub-data includes complete character sub-data corresponding to the second complete sub-data and incomplete character sub-data corresponding to the incomplete sub-data;

[0029] Based on the built-in grammatical structure, the complete character sub-data and the incomplete character sub-data are compared with each other, the grammatical structure of the incomplete character sub-data is corrected, and the semantic integrity of the corrected incomplete character sub-data is judged to obtain the recognized character data with complete semantics;

[0030] Determine the SMS category for the recognized text data with complete semantics.

[0031] Preferably, the security of the text data is judged based on the SMS category of the text data;

[0032] If the SMS is classified as advertising, the SMS content is searched for purchase channels to determine the purchase channels and to determine the safety of the purchase channels;

[0033] If the purchase channel is a secure platform, the SMS content is determined to be secure and the security value of the SMS content is recorded as 1. If the purchase channel is an unsecure platform, the consumption amount of the product information is determined based on the SMS content and compared with the built-in security amount.

[0034] If the consumption amount is greater than or equal to the built-in safety amount, the safety value of the SMS content is determined to be 0; if the consumption amount is less than the built-in safety amount, the ratio of the consumption amount to the safety amount is calculated, and the difference between the safety value in the safety state and the ratio is recorded as the safety value of the SMS content;

[0035] If the SMS category is junk, the security value of the SMS content is 1;

[0036] If the SMS category is a misleading category, the core content of the text data is obtained, and the dangerousness of the core content is judged to determine the safety value of the SMS content. The core content may be misleading link clicks, misleading gambling, etc., and the link data and misleading gambling correspond to different safety values.

[0037] The safety value corresponding to each SMS category is used as the risk value of the text data.

[0038] Preferably, based on the SMS category, if the SMS category is pure link data, the account information of the SMS content is obtained, and the SMS content received by the user is read according to the account information to obtain the SMS content set;

[0039] Determine the SMS category in the SMS content set. If the SMS content set only includes link data, the sum of the security value of the link data and the credibility value of the SMS content is used as the risk index of the SMS information.

[0040] If the SMS content set includes text data, then performing correlation matching on the text data and the link data to determine correlation data between the text data and the link data;

[0041] Based on the SMS category of the text data, the security value of the text data is obtained, and based on the security value, credibility value and correlation data, the security of the SMS information is judged to determine the danger index of the SMS information.

[0042] Preferably, the sender information of the SMS message is read to determine whether the sender is a known person;

[0043] If the sender is determined to be a known person, the SMS exchange data between the user and the sender is obtained, and the frequency of information communication between the user and the sender is determined based on the SMS exchange data;

[0044] Match the frequency of information communication with the built-in credibility table to obtain the sender's credibility value;

[0045] If the sender is determined to be an unknown person, key information is read from the text data in the SMS message to determine whether key information of the person exists in the text data;

[0046] If it is determined that there is key information of the person in the text data, the key information of the person is matched with the built-in credibility table to determine the credibility value of the sender; wherein the built-in credibility table includes credibility values ​​under different information communication frequencies and credibility values ​​corresponding to different key information.

[0047] If it is determined that the text data does not contain key information about the person, the built-in credibility value of the unknown person is used as the credibility value of the sender.

[0048] Preferably, historical SMS messages received by the user are obtained, and link data is screened for the historical SMS messages to obtain SMS messages containing link data; because when judging the user's susceptibility index, plain text SMS messages do not have any fraudulent means, so it is impossible to defraud the user, and link data has communication and virus functions, so it is necessary to judge the SMS messages with link data, thereby ensuring the accuracy of the susceptibility index obtained based on the analysis of the SMS messages.

[0049] Based on the SMS message containing the link data, obtaining the user's historical operation data corresponding to the SMS message;

[0050] Perform link click operation screening on the user's historical operation data to obtain SMS messages with link click operations, and mark the SMS messages as operation SMS messages;

[0051] Classify the operation text messages into text message categories to obtain first classification results for each text message category;

[0052] Classifying the SMS messages containing the link data into SMS categories to obtain second classification results for each SMS category;

[0053] The first division result and the second division result are calculated according to the SMS category to obtain the ratio data, and the ratio data is used as the user's fraud index for each SMS category.

[0054] In a second aspect, the present application provides a communication text information conversion system, which adopts the following technical solution:

[0055] A communication text information conversion system, comprising: a short message acquisition module, a category judgment module, a danger judgment module and a short message interception module;

[0056] The SMS acquisition module acquires the SMS information received by the user and identifies the components of the SMS information to obtain information component data; the information component data includes text data and link data;

[0057] The category judgment module judges whether the information component data contains text data. If the information component data contains text data, the text data is converted into text information and analyzed to determine the category of the short message; if the information component data does not contain text data, the short message category of the short message is determined to be a pure link category;

[0058] The risk judgment module reads key information of the SMS message and matches the key information with the built-in credibility table to obtain a credibility value; judges the text data in the message component data to determine the danger of the text data; judges the link data in the message component data to determine the danger of the link data; determines the danger index of the SMS message based on the credibility value of the SMS message, the danger of the text data and the danger of the link data;

[0059] The SMS interception module matches the user's spoofing index based on the SMS category of the SMS information to obtain the user's spoofing index in the corresponding SMS category; compares the danger index of the SMS information with the spoofing index in the corresponding SMS category, marks the SMS information whose danger index is less than the spoofing index in the corresponding SMS category as safe SMS and presents it, and marks the SMS information whose danger index is greater than or equal to the spoofing index in the corresponding SMS category as dangerous SMS and intercepts it.

[0060] In summary, the present application includes at least one of the following beneficial technical effects:

[0061] 1. By identifying the components of the SMS received by the user, and separately analyzing the identified text data and link data, the dangerousness of the text data and link data is determined, and then the text data and link data are jointly judged according to the components of the SMS. When the SMS contains only text data, the dangerousness of the text data is used as the dangerousness of the SMS. When the SMS contains only link data, the dangerousness of the link data is used as the dangerousness of the SMS. When the SMS contains both text data and link data, the comprehensive judgment result of the text data and link data is used as the dangerousness of the SMS, so that the judgment result of the dangerousness of the SMS is more accurate. Then, through the user's fraud index for different categories of SMS, the dangerousness of this type of information to the user is determined, thereby ensuring the effective interception of SMS, reducing the probability of false interception, and improving the interception effect of SMS;

[0062] 2. By performing semantic conversion on text data, the semantic integrity of the text data is judged for the first time, ensuring the accuracy of subsequent judgment results. When the text data is determined to be complete, the core content of the text data is judged and matched with the advertising category, junk SMS category and inducement category, so as to clarify the category of the SMS information and make targeted judgments on the SMS according to the category, providing basic data for SMS recognition accuracy. When it is determined that the text data is not complete, the syllable data of the text data is read and converted, so as to make the recognition text data generated according to the syllable data more accurate, ensuring the system's correct understanding of the SMS, thereby improving the accuracy of SMS judgment;

[0063] 3. With the help of grammatical structure and semantic integrity, the recognized text data is split into sentences according to semantic integrity to clarify semantically incomplete sentences. By comparing semantically complete sentences with semantically incomplete sentences, the semantically incomplete sentences are supplemented to make the semantically incomplete sentences complete, thereby making the judgment result more accurate when judging the SMS category of the recognized text data, and improving the accuracy of identifying the dangerousness of SMS. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is a flowchart of the steps of the communication text information conversion method of this embodiment;

[0065] Figure 2 This is a module block diagram of the communication text information conversion system of this embodiment.

[0066] Figure numerals: 1. SMS acquisition module; 2. Category judgment module; 3. Danger judgment module; 4. SMS interception module. DETAILED DESCRIPTION

[0067] The following is combined with Figure 1-Figure 2 This application is described in further detail.

[0068] The embodiments of the present application disclose a communication text information conversion method and system.

[0069] Example: Figure 1 As shown, the present invention provides a communication text information conversion method, comprising:

[0070] S100, obtaining the SMS information received by the user, and identifying the components of the SMS information to obtain information component data; the information component data includes text data and link data; because when the user receives the SMS, the content of the SMS may include plain text data and link data containing a website link, the received SMS is judged separately to make the judgment result of the SMS more accurate.

[0071] S200, determining whether the information component data contains text data. If the information component data contains text data, performing text information conversion analysis on the text data to determine the SMS category. By performing category determination on the received SMS content, corresponding determination measures can be taken according to different SMS categories, thereby making the determination result more accurate.

[0072] S300, if the information component data does not include text data, determining that the SMS category of the SMS message is a pure link category;

[0073] S400, reading key information of the SMS message, and matching the key information with the built-in credibility table to obtain a credibility value; wherein the information key words include the name of the SMS sender and the relationship between the sender and the user.

[0074] S500, judging the text data in the information component data to determine the danger of the text data;

[0075] S600, judging the link data in the information component data to determine the danger of the link data; by judging the text data and the link data separately, the corresponding judgment results are more accurate, and thus the judgment result of the entire text message content is more accurate. When judging the security of the link data, the security of the link is judged by using the deep learning algorithm in the prior art to determine the danger of the link data.

[0076] S700, based on the credibility value of the SMS information, the danger of the text data and the danger of the link data, determine the danger index of the SMS information; for example, when the SMS content is only text data, since there is no way to be deceived, its own danger is low, and if the SMS content is only link data, since the sender of the message is unknown and the security of the link is unknown, the click rate on the link is low, and thus the danger is low, and when the SMS content contains both text data and link data, there may be a possibility of guiding the user to the link data through the text data, thereby increasing the probability of the user clicking on the unknown link, and thus increasing the danger, so it is necessary to strictly evaluate this type of SMS to ensure the safety of the user and reduce the probability of being deceived.

[0077] S800, based on the SMS category of the SMS information, the user's fraud index is matched to obtain the user's fraud index in the corresponding SMS category; the same user has different trust in SMS of different categories, so the fraud index of each user in different SMS is judged, so that the result obtained by analysis and judgment is more accurate.

[0078] S900, compare the danger index of the SMS message with the fraud index of the corresponding SMS category, mark the SMS message with a danger index less than the fraud index of the corresponding SMS category as a safe SMS and present it, mark the SMS message with a danger index greater than or equal to the fraud index of the corresponding SMS category as a dangerous SMS and intercept it. When the danger index of the SMS message is greater than the fraud index, it indicates that the user has a higher fraud rate for the SMS message, so it needs to be intercepted to reduce the fraud rate, and when the danger index of the SMS message is less than the fraud index, it indicates that the user has a strong resistance to this type of SMS, so in order to prevent misjudgment of the SMS content, the user can be used to screen it by himself, thereby reducing the fraud rate of the user and ensuring that the user checks the information in time to avoid misinterpretation.

[0079] In this embodiment, the components of the text messages received by the user are identified, and the identified text data and link data are analyzed separately to determine the danger of the text data and link data, and then the text data and link data are jointly judged according to the component composition of the text message. When the text message contains only text data, the danger of the text data is used as the danger of the text message. When the text message contains only link data, the danger of the link data is used as the danger of the text message. When the text message contains both text data and link data, the comprehensive judgment result of the text data and link data is used as the danger of the text message, thereby making the judgment result of the danger of the text message more accurate. Then, the danger of this type of information to the user is determined through the user's susceptibility index to different types of text messages, thereby ensuring the effective interception of text messages, reducing the probability of false interception, and improving the interception effect of text messages.

[0080] In step S200, if the information component data includes text data, the text data is subjected to text information conversion analysis to determine the SMS category, including the following steps:

[0081] S210, semantically understand the text data to obtain first semantic data, and make a semantic integrity judgment on the first semantic data; when semantically understanding the text data, existing technology can be used: a natural language analysis algorithm is used to analyze and understand the text data, and then an accurate judgment is made on whether the semantics of the text data is complete.

[0082] S220, if the first semantic data is complete, the core content of the text data is judged to determine whether the core content is product information; when the semantics of the text data is complete, it indicates that the expression content of the text message is clear, and the result obtained during the analysis and judgment is more accurate and effective.

[0083] S230, if the core content is product information, determining the SMS category is advertising and promotion;

[0084] S240, if the core content is not product information, then perform sensitive word matching on the text data to determine whether sensitive words exist in the text data; wherein sensitive words include false statements, insulting words, etc.

[0085] S250, if there are sensitive words in the text data, determining that the text message is a spam message;

[0086] S260, if there are no sensitive words in the text data, the text data is judged to be inducement based on the first semantic data; the text data is identified to be inducement based on the natural language understanding technology, and when inducement is identified, it is determined to be inducement, otherwise it is determined not to be inducement.

[0087] S270, if the text data is inductive to the user's behavior, then determine that the SMS category is an inductive category;

[0088] S280, if the first semantic data is not complete, then obtain the syllable data of the text data, generate the corresponding voice data according to the syllable data, and perform text conversion on the voice data to obtain the recognized text data; wherein, the syllable data is the pronunciation of the text, that is, the initials, finals, tones, etc. When it is determined that the first semantic data is not complete, it indicates that the text data corresponding to the first semantic data has situations such as typos, rare characters, homophones with different characters, and missing characters. Therefore, by obtaining the syllable data of the text data and then regenerating the text according to the syllable data, situations such as homophones with different characters are excluded, improving the accuracy of the analysis result. For example, when there are homophonic words in the text data, it will cause the first semantic data understood by the system to be incomplete. Therefore, by using the same pronunciation of the homophonic words for homophonic word conversion, the corresponding recognized text data that can be understood by the system is obtained. For example: the text data is "Help users like their works to get on the hot list and increase followers", and in the semantic understanding, "men" cannot be jointly understood with other texts, resulting in the first semantic data being incomplete. However, after the syllable data conversion, what is obtained is "Help users like their works to get on the hot topic and increase followers", which can be fully understood during the semantic understanding process.

[0089] S290, perform semantic understanding on the recognized text data to obtain the second semantic data;

[0090] S2100, perform a semantic integrity judgment on the second semantic data to determine the integrity of the second semantic data, and determine the SMS category of the recognized text data according to the integrity of the second semantic data.

[0091] In this embodiment, during the process of confirming the SMS category according to the text data, by making the first judgment on the semantic integrity of the text data, the accuracy of the subsequent judgment results is ensured. When it is determined that the text data is complete, by judging the central content of the text data and matching the central content with the advertising promotion category, junk SMS category, and inductive category, the category of the SMS information is clarified, and then the SMS can be judged specifically according to the category, providing basic data for the accuracy judgment of SMS recognition. When it is determined that the text data is not complete, by reading and converting the syllable data of the text data, the recognized text data generated according to the syllable data is made more accurate, ensuring the correct understanding of the SMS by the system, and thus improving the judgment accuracy of the SMS.

[0092] In step S2100, the second semantic data is subjected to semantic integrity judgment to determine the integrity of the second semantic data, and the SMS category of the recognized text data is determined according to the integrity of the second semantic data, including the following steps:

[0093] S2101, perform semantic integrity judgment on the second semantic data. If the second semantic data is complete, determine the SMS category of the recognized text data. When it is determined that the second semantic data is complete, it indicates that the recognized text data corresponding to the second semantic data can accurately express the information.

[0094] S2102, if the second semantic data is not complete, the second semantic data is divided according to the completeness of the sentence to obtain second semantic sub-data with sentence order; the second semantic sub-data includes second complete sub-data with complete semantics and second incomplete sub-data without complete sentences; since a text message may contain multiple sentences, separate semantic judgments are performed on the multiple sentences to further clarify the existence of incomplete sentences.

[0095] S2103, splitting the recognized character data according to the second semantic sub-data to obtain recognized character sub-data corresponding to the second semantic sub-data; the recognized character sub-data includes complete character sub-data corresponding to the second complete sub-data and incomplete character sub-data corresponding to the incomplete sub-data;

[0096] S2104, based on the built-in grammatical structure, the complete text sub-data and the incomplete text sub-data are compared with each other, the grammatical structure of the incomplete text sub-data is corrected, and the semantic integrity of the corrected incomplete text sub-data is judged to obtain the recognized text data with complete semantics; by comparing the grammatical structure with the incomplete text sub-data, the incomplete grammatical structure of the incomplete text sub-data is determined, and then the complete text sub-data is searched for the same grammatical structure to obtain text that can supplement the incomplete grammatical structure of the incomplete text sub-data, and then the incomplete text sub-data is semantically judged to obtain text that can meet the semantic integrity of the incomplete text sub-data, and then the complete text sub-data is combined to make an overall semantic integrity judgment to obtain the recognized text data that can finally meet the needs.

[0097] S2105, determining the SMS category of the recognized text data with complete semantics.

[0098] In this embodiment, when the second semantic data is subjected to a common judgment on semantic completeness, when the second semantic data is incomplete, the recognized text data is split into sentences according to semantic completeness to clarify the semantically incomplete sentences, and the semantically complete sentences are compared with the semantically incomplete sentences, and the semantically incomplete sentences are supplemented to make the semantically incomplete sentences complete, thereby making the judgment result more accurate when the recognized text data is subjected to a text message category judgment, thereby making the judgment result of the dangerousness of the text message more accurate, and improving the effect of text message interception.

[0099] In step S400, the key information of the SMS message is read and matched with the built-in credibility table to obtain the credibility value, including the following steps:

[0100] S410, read the sender information of the SMS message to determine whether the sender is a known person; wherein the known person is the mobile phone number that the user has marked with the sender information. For example, mobile phone number a sends a SMS message to the user, and the user marks the person information of mobile phone number a as A, which indicates that mobile phone number a is a known person. By judging the sender's personal information, the relationship between the sender and the user is clarified. When the sender is determined to be a known person, it indicates that the user has a certain understanding of the sender, thereby establishing a certain trust basis between the user and the sender.

[0101] S420, if it is determined that the sender is a known person, the SMS exchange data between the user and the sender is obtained, and the frequency of information communication between the user and the sender is determined based on the SMS exchange data; when the sender is determined to be a known person, since the chat frequency between the user and the sender is higher, the corresponding contact between the two parties is closer, the closer the relationship is, the higher the trust between the sender and the user, and thus the higher the credibility value.

[0102] S430, matching the information communication frequency with the built-in credibility table to obtain the credibility value of the sender;

[0103] S440, if it is determined that the sender is an unknown person, the text data in the SMS message is read for key information to determine whether the text data contains key information of the person; and when the sender is an unknown person, it means that the sender and the user are not acquainted, and further judgment is required to clarify the credibility of the SMS sent by the sender. For example, if a and b know each other, the SMS sent by a to b is more credible.

[0104] S450, if it is determined that there is key information of the person in the text data, the key information of the person is matched with the built-in credibility table to determine the credibility value of the sender; wherein the built-in credibility table includes credibility values ​​under different information communication frequencies and credibility values ​​corresponding to different key information.

[0105] S460: If it is determined that the text data does not contain key information of the person, the built-in credibility value of the unknown person is used as the credibility value of the sender.

[0106] In this embodiment, the personal information of the sender of the SMS message is read to make a first judgment on the sender. When it is determined to be an unknown person, a second judgment is made on the text data of the SMS message to further clarify the personal information of the sender, and then clarify the relationship between the sender and the user, so that the SMS message obtained based on the frequency of information communication and key information of the person is more credible.

[0107] In step S500, the text data in the information component data is judged to determine the danger of the text data, including the following steps:

[0108] S510, judging the security of the text data based on the SMS category of the text data; when performing security analysis on the text data, taking different analysis and judgment measures according to different SMS categories, thereby improving the accuracy of the security judgment on the text data.

[0109] S520, if the SMS category is advertising and promotion, the purchase channel is searched for in the SMS content to determine the purchase channel and make a security assessment on the purchase channel; when it is determined that the SMS category is advertising and promotion, only the security assessment on the channel platform of advertising and promotion is required to ensure that the user's consumer rights can be effectively protected when purchasing promotional products.

[0110] S530, if the purchase channel is a secure platform, the SMS content is determined to be secure, and the security value of the SMS content is recorded as 1; when the purchase channel is determined to be secure, it indicates that the user's consumer rights can be protected, so the SMS is secure. At the same time, the security of the SMS is not equal to the validity of the SMS, so the secure advertising promotion is intercepted or presented by using the existing SMS interception rules.

[0111] S540, if the purchase channel is a non-secure platform, determine the consumption amount of the product information according to the content of the SMS, and compare the consumption amount with the built-in security amount;

[0112] S550, if the consumption amount is greater than or equal to the built-in security amount, the security value of the text message content is determined to be 0; if the consumption amount is less than the built-in security amount, the ratio data of the consumption amount and the security amount is calculated, and the difference between the security value in the safe state and the ratio data is recorded as the security value of the text message content; when the purchase channel is determined to be a non-safe platform, it means that the user's consumer rights are not guaranteed, so it is necessary to evaluate and protect the consumer's financial resources, thereby ensuring that the user's losses when their rights are not guaranteed after consumption are minimized, thereby improving the user's property security.

[0113] S560, if the SMS is classified as spam, the security value of the SMS content is determined to be 1; when the SMS is classified as spam, it indicates that the SMS is safe but invalid, so it can be intercepted or presented according to the existing spam interception rules.

[0114] S570, if the SMS category is an inducement category, the core content of the text data is obtained, and the dangerousness of the core content is judged to determine the safety value of the SMS content. The core content may be inducement to click on a link, inducement to gambling, etc., and the link data and inducement to gambling correspond to different safety values. When the SMS category is determined to be an inducement category, the dangerousness of the core content of the text data is judged. For example, if the core content of the text data is gambling, fraud, etc., the corresponding dangerousness is determined according to the different core contents, and then the corresponding safety value is obtained.

[0115] S580: Using the safety value corresponding to each SMS category as the risk value of the text data.

[0116] In this embodiment, different analysis and judgment methods are adopted for different categories of SMS messages, so that the final analysis results are made more accurate. For advertising and promotion categories, the security of the SMS message is determined by making a security judgment on the purchase channel. For inducement categories, the security of the SMS content is determined by determining the core content and then performing security matching on the core content.

[0117] In step S700, based on the credibility value of the SMS message, the danger of the text data and the danger of the link data, determining the danger index of the SMS message includes the following steps:

[0118] S710, based on the SMS category, if the SMS category is pure link data, obtain the account information of the SMS content sent, read the SMS content received by the user according to the account information, and obtain the SMS content set;

[0119] S720, determine the SMS category in the SMS content set. If the SMS content set only includes link data, the sum of the security value of the link data and the credibility value of the SMS content is used as the danger index of the SMS information. Since the security of the connection data only includes two situations, dangerous and safe, the dangerous state is judged as 0 and the safe state is judged as 1. Therefore, when the link data is in a safe state and the credibility of the user is also x, where x is less than or equal to 1, the sum of the security value of the connection data and the credibility value of the SMS content is less than 1.

[0120] S730, if the SMS content set includes text data, then performing correlation matching on the text data and the link data to determine correlation data between the text data and the link data;

[0121] S740, based on the SMS category of the text data, obtain the security value of the text data, and based on the security value, the credibility value and the correlation data, judge the security of the SMS information to determine the danger index of the SMS information.

[0122] In this embodiment, the text messages sent by the same text message sender are counted to ensure the integrity of the analysis results, thereby reducing misjudgment of text message information. The text message sender is first judged to determine the credibility of the text message, and then the text message category of the text message content is judged, and corresponding calculations are performed according to different text message categories. The corresponding risk index is calculated according to different text message categories. When the text message to be evaluated is pure link data, and the same text message sender as the text message to be evaluated includes other text data, the correlation between the text data and the link data is determined to determine whether the other text data provides user guidance or explanation for the link data, thereby making the risk judgment of the link data to be evaluated more accurate and improving the text message interception effect.

[0123] In step S800, the user's fraud index includes the following steps:

[0124] S810, obtain historical SMS messages received by the user, and filter the link data of the historical SMS messages to filter out SMS messages containing link data; when judging the user's susceptibility index, plain text SMS messages do not have any fraudulent means, so it is impossible to defraud the user, and link data has communication and virus functions, so it is necessary to judge the SMS messages with link data, thereby ensuring the accuracy of the susceptibility index obtained based on the analysis of the SMS messages.

[0125] S820, based on the SMS message containing the link data, obtaining user historical operation data corresponding to the SMS message;

[0126] S830, filter the user's historical operation data for link click operations, obtain SMS messages with link click operations, and mark the SMS messages as operation SMS messages; wherein the operation SMS messages are all SMS messages containing link click operations.

[0127] S840, classify the operation text messages into text message categories to obtain first classification results for each text message category; wherein the first classification results are text messages containing link click operations in the same text message category.

[0128] S850, classifying the SMS information containing the link data into SMS categories to obtain second classification results for each SMS category; wherein the second classification results are SMS messages containing the link data in the same SMS category.

[0129] S860, calculate the ratio of the first division result to the second division result according to the SMS category to obtain ratio data, and use the ratio data as the user's fraud index for each SMS category. The fraud index is the proportion of SMS with link click operations in the same SMS category to all SMS containing link data.

[0130] In this embodiment, the link data of all SMS messages received by the user is screened to ensure that the screened SMS messages have the validity of calculating the fraud index, and then the operation data of the corresponding SMS messages are screened to determine the number of SMS messages with fraudulent behavior. The SMS message data with fraudulent behavior and the number of SMS messages containing link data are compared to the corresponding SMS categories to determine the probability of users clicking on link data in different SMS categories, and then determine the possibility of users being deceived by the link data in the SMS messages, so that the corresponding fraud index is more accurate and effective, thereby improving the danger of SMS messages judged according to the fraud index, and making the system more accurate in intercepting SMS messages.

[0131] Based on the description of the above communication text information conversion method embodiment, the embodiment of the present invention also discloses a communication text information conversion system:

[0132] like Figure 2 As shown, a communication text information conversion system, by applying the above communication text information conversion method, includes: a text message acquisition module 1, a category judgment module 2, a danger judgment module 3 and a text message interception module 4;

[0133] The SMS acquisition module 1 acquires the SMS information received by the user and identifies the components of the SMS information to obtain information component data; the information component data includes text data and link data;

[0134] Category judgment module 2, judges whether the information component data contains text data, if the information component data contains text data, performs text information conversion analysis on the text data to determine the SMS category; if the information component data does not contain text data, determines that the SMS category of the SMS information is a pure link category;

[0135] The danger judgment module 3 reads the key information of the SMS message and matches the key information with the built-in credibility table to obtain the credibility value; judges the text data in the message component data to determine the danger of the text data; judges the link data in the message component data to determine the danger of the link data; determines the danger index of the SMS message based on the credibility value of the SMS message, the danger of the text data and the danger of the link data;

[0136] The SMS interception module 4 matches the user's fraud index based on the SMS category of the SMS information to obtain the user's fraud index in the corresponding SMS category; compares the danger index of the SMS information with the fraud index in the corresponding SMS category, marks the SMS information whose danger index is less than the fraud index in the corresponding SMS category as safe SMS and presents it, marks the SMS information whose danger index is greater than or equal to the fraud index in the corresponding SMS category as dangerous SMS and intercepts it.

[0137] Compared with the existing communication text information conversion method and system, the present invention improves the recognition accuracy of harassing text messages.

[0138] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A communication text information conversion method, characterized in that: include: Obtain the SMS information received by the user, and identify the components of the SMS information to obtain information component data; The information component data includes text data and link data; Determine whether the information component data contains text data, and if the information component data contains text data, perform text information conversion analysis on the text data to determine the SMS category; If the information component data does not contain text data, the SMS category of the SMS message is determined to be a pure link category; Read the key information of the SMS message and match the key information with the built-in credibility table to obtain the credibility value; Judging the text data in the information component data to determine the danger of the text data; Judging the linked data in the information component data to determine the danger of the linked data; Determine the risk index of the SMS message based on the credibility value of the SMS message, the risk of the text data, and the risk of the link data; Based on the SMS category of the SMS information, the user's fraud index is matched to obtain the user's fraud index in the corresponding SMS category; Compare the danger index of the SMS message with the fraud index of the corresponding SMS category, mark the SMS message with a danger index less than the fraud index of the corresponding SMS category as a safe SMS and present it, and mark the SMS message with a danger index greater than or equal to the fraud index of the corresponding SMS category as a dangerous SMS and intercept it.

2. A communication text information conversion method according to claim 1, characterized in that: If the information component data includes text data, the text data is subjected to text information conversion analysis to determine the SMS category, specifically: Performing semantic understanding on the text data to obtain first semantic data, and performing semantic integrity judgment on the first semantic data; If the first semantic data is complete, then performing a core content judgment on the text data to determine whether the core content is product information; If the core content is product information, the SMS category is determined to be advertising and promotion; If the central content is not product information, the text data is matched with sensitive words to determine whether sensitive words exist in the text data; If there are sensitive words in the text data, the SMS is classified as spam; If there are no sensitive words in the text data, performing an inductive judgment on the text data according to the first semantic data; If the text data is misleading to the user's behavior, the SMS category is determined to be misleading; If the first semantic data is not complete, obtaining syllable data of the text data, generating corresponding voice data according to the syllable data, and converting the voice data into text to obtain recognized text data; Performing semantic understanding on the recognized text data to obtain second semantic data; A semantic integrity judgment is performed on the second semantic data to determine the integrity of the second semantic data, and the SMS category of the recognized text data is determined based on the integrity of the second semantic data.

3. A communication text information conversion method according to claim 2, characterized in that: The performing of semantic integrity judgment on the second semantic data to determine the integrity of the second semantic data, and determining the SMS category of the recognized text data according to the integrity of the second semantic data, is specifically as follows: Performing semantic integrity judgment on the second semantic data, and if the second semantic data is complete, determining the SMS category of the recognized text data; If the second semantic data is not complete, the second semantic data is divided according to the completeness of the sentence to obtain second semantic sub-data with a sentence order; the second semantic sub-data includes second complete sub-data with complete semantics and second incomplete sub-data without a complete sentence; The recognized character data is split according to the second semantic sub-data to obtain recognized character sub-data corresponding to the second semantic sub-data; the recognized character sub-data includes complete character sub-data corresponding to the second complete sub-data and incomplete character sub-data corresponding to the incomplete sub-data; Based on the built-in grammatical structure, the complete character sub-data and the incomplete character sub-data are compared with each other, the grammatical structure of the incomplete character sub-data is corrected, and the semantic integrity of the corrected incomplete character sub-data is judged to obtain the recognized character data with complete semantics; Determine the SMS category for the recognized text data with complete semantics.

4. A communication text information conversion method according to claim 3, characterized in that: The text data in the information component data is judged to determine the danger of the text data, specifically: Based on the SMS category of the text data, the security of the text data is judged; If the SMS is classified as advertising, the SMS content is searched for purchase channels to determine the purchase channels and to determine the safety of the purchase channels; If the purchase channel is a secure platform, the SMS content is determined to be secure, and the security value of the SMS content is recorded as 1; If the purchase is made through a non-secure platform, the consumption amount of the product information is determined based on the content of the SMS, and the consumption amount is compared with the built-in security amount; If the consumption amount is greater than or equal to the built-in security amount, the security value of the SMS content is determined to be 0; If the consumption amount is less than the built-in safety amount, the ratio of the consumption amount to the safety amount is calculated, and the difference between the safety value in the safety state and the ratio is recorded as the safety value of the SMS content; If the SMS is classified as spam, the security value of the SMS content is 1; If the SMS category is misleading, the core content of the text data is obtained, and the dangerousness of the core content is judged to determine the safety value of the SMS content; The safety value corresponding to each SMS category is used as the risk value of the text data.

5. A communication text information conversion method according to claim 4, characterized in that: The risk index of the SMS message is determined based on the credibility value of the SMS message, the risk of the text data and the risk of the link data, specifically: Based on the SMS category, if the SMS category is pure link data, the account information of the SMS content is obtained, and the SMS content received by the user is read according to the account information to obtain the SMS content set; Determine the SMS category in the SMS content set. If the SMS content set only includes link data, the sum of the security value of the link data and the credibility value of the SMS content is used as the risk index of the SMS information. If the SMS content set includes text data, then performing correlation matching on the text data and the link data to determine correlation data between the text data and the link data; Based on the SMS category of the text data, the security value of the text data is obtained, and based on the security value, credibility value and correlation data, the security of the SMS information is judged to determine the danger index of the SMS information.

6. A communication text information conversion method according to claim 1, characterized in that: The key information of the SMS message is read, and the key information is matched with the built-in credibility table to obtain the credibility value, which is specifically: Read the sender information of the SMS message to determine whether the sender is a known person; If the sender is determined to be a known person, the SMS exchange data between the user and the sender is obtained, and the frequency of information communication between the user and the sender is determined based on the SMS exchange data; Match the frequency of information communication with the built-in credibility table to obtain the sender's credibility value; If the sender is determined to be an unknown person, key information is read from the text data in the SMS message to determine whether key information of the person exists in the text data; If it is determined that the text data contains key information of the person, the key information of the person is matched with the built-in credibility table to determine the credibility value of the sender; If it is determined that the text data does not contain key information about the person, the built-in credibility value of the unknown person is used as the credibility value of the sender.

7. A communication text information conversion method according to claim 1, characterized in that: The user's fraud index is specifically: Obtain historical SMS messages received by the user, and filter the historical SMS messages for link data to filter out SMS messages containing link data; Based on the SMS message containing the link data, obtaining the user's historical operation data corresponding to the SMS message; Perform link click operation screening on the user's historical operation data to obtain SMS messages with link click operations, and mark the SMS messages as operation SMS messages; Classify the operation text messages into text message categories to obtain first classification results for each text message category; Classifying the SMS messages containing the link data into SMS categories to obtain second classification results for each SMS category; The first division result and the second division result are calculated according to the SMS category to obtain the ratio data, and the ratio data is used as the user's fraud index for each SMS category.

8. A communication text information conversion system, characterized in that: The system is used to implement a communication text information conversion method as described in any one of claims 1 to 7: comprising: a text message acquisition module, a category judgment module, a danger judgment module and a text message interception module; The SMS acquisition module acquires the SMS information received by the user and identifies the components of the SMS information to obtain information component data; the information component data includes text data and link data; The category judgment module judges whether the information component data contains text data. If the information component data contains text data, the text data is converted into text information and analyzed to determine the category of the short message; if the information component data does not contain text data, the short message category of the short message is determined to be a pure link category; The risk judgment module reads key information of the SMS message and matches the key information with the built-in credibility table to obtain a credibility value; judges the text data in the message component data to determine the danger of the text data; judges the link data in the message component data to determine the danger of the link data; determines the danger index of the SMS message based on the credibility value of the SMS message, the danger of the text data and the danger of the link data; The SMS interception module matches the user's spoofing index based on the SMS category of the SMS information to obtain the user's spoofing index in the corresponding SMS category; compares the danger index of the SMS information with the spoofing index in the corresponding SMS category, marks the SMS information whose danger index is less than the spoofing index in the corresponding SMS category as safe SMS and presents it, and marks the SMS information whose danger index is greater than or equal to the spoofing index in the corresponding SMS category as dangerous SMS and intercepts it.

Images