Electronic information identification method and device, electronic equipment, storage medium and computer program product
Patent Information
- Application Number
- CN202411642163.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-18
- Publication Date
- 2025-04-25
AI Technical Summary
In the prior art, when manually identifying electronic information, the efficiency is inefficient and insufficient accuracy, resulting in a large amount of labor consumption and missed judgments or misjudgments.
An abnormal information recognition model constructed based on Bayesian formula is adopted, and the abnormality probability information caused by the electronic information to be recognized is obtained by using word segmentation processing and inputting it into the abnormality information recognition model, and the abnormality probability information of each word causes abnormality of electronic information is finally determined.
It realizes the abnormality of electronic information quickly and accurately identifying it, and improves the efficiency of electronic information recognition.
Smart Images
Figure CN119884373A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of communication technology, and in particular to an electronic information identification method, device, electronic device, storage medium and computer program product. Background Art
[0002] Currently, abnormal electronic information such as spam emails and spam text messages are identified by manual reading and judgment. However, manual identification consumes a lot of manpower and is prone to lack of accuracy due to missed or misjudgment, which makes the current electronic information identification inefficient. Summary of the invention
[0003] The present application aims to solve at least one of the technical problems existing in the related art. To this end, the present application proposes an electronic information identification method, device, electronic device, storage medium and computer program product to solve the problem that the current electronic information identification by manual labor consumes a lot of manpower and is not accurate enough, so as to improve the recognition efficiency of electronic information.
[0004] According to the first aspect of the present application, the electronic information identification method includes: Performing word segmentation processing on the electronic information to be identified to obtain at least one word; Input each word into the abnormal information recognition model to obtain abnormal probability information of each word output by the abnormal information recognition model causing the abnormality of the electronic information to be recognized; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula; The abnormality identification result of the electronic information to be identified is determined according to each abnormality probability information.
[0005] According to an embodiment of the present application, the abnormal probability information includes at least one abnormal type and an abnormal probability corresponding to the corresponding abnormal type; the abnormal identification result of the electronic information to be identified is determined according to each abnormal probability information, including: Adding the abnormality probabilities of the same abnormality type of each word to obtain at least one summed probability; Compare each summed probability with a preset probability threshold to obtain a comparison result; The abnormality identification result of the electronic information to be identified is determined according to the comparison result.
[0006] According to an embodiment of the present application, determining the abnormality identification result of the electronic information to be identified according to the comparison result includes: If any sum probability in the comparison result is greater than the preset probability threshold, the electronic information to be identified is determined to be abnormal information, and the abnormal type corresponding to the sum probability greater than the preset probability threshold is determined as the abnormal type of the abnormal information.
[0007] According to an embodiment of the present application, the preset probability threshold is adjusted based on a preset initial probability threshold.
[0008] According to an embodiment of the present application, the word segmentation processing of the electronic information to be identified to obtain at least one vocabulary includes: The electronic information to be identified is input into a word segmentation model to obtain at least one word output by the word segmentation model; wherein the word segmentation model is trained based on sample electronic information and its word segmentation labels.
[0009] According to an embodiment of the present application, after determining the abnormality identification result of the electronic information to be identified according to each abnormality probability information, it also includes: A processing scheme for the electronic information to be identified is determined according to the abnormal identification result of the electronic information to be identified.
[0010] According to the second aspect of the present application, an electronic information identification device includes: A word segmentation module, used for performing word segmentation processing on the electronic information to be identified to obtain at least one word; The recognition module is used to input each word into the abnormal information recognition model to obtain the abnormal probability information that each word output by the abnormal information recognition model causes the abnormality of the electronic information to be recognized; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula; The determination module is used to determine the abnormality identification result of the electronic information to be identified according to each abnormality probability information.
[0011] According to an electronic device of an embodiment of the third aspect of the present application, the electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the electronic information identification method described in any one of the above methods is implemented.
[0012] According to the storage medium of the fourth aspect of the present application, the storage medium is a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements any of the electronic information identification methods described above.
[0013] According to the computer program product of the fifth aspect of the present application, the computer program includes a computer program, which, when executed by a processor, implements any of the electronic information identification methods described above.
[0014] The above one or more technical solutions in the embodiments of the present application have at least the following technical effects: By constructing a preset model in advance through the Bayesian formula, and training the preset model through sample electronic information and its information labels to obtain an abnormal information recognition model, after the electronic information to be recognized is segmented to obtain at least one word, each word is input into the abnormal information recognition model, and the abnormal probability information of each word output by the abnormal information recognition model causing the electronic information to be recognized to be abnormal can be obtained quickly and accurately, and then according to each abnormal probability information, the abnormal recognition result of the electronic information to be recognized can be quickly and accurately determined, thereby improving the recognition efficiency of electronic information.
[0015] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] Figure 1 It is a flow chart of the electronic information identification method provided in the embodiment of the present application.
[0018] Figure 2 It is a structural schematic diagram of the electronic device provided by this application. DETAILED DESCRIPTION
[0019] The following is a further detailed description of the implementation of the present application in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present application but cannot be used to limit the scope of the present application.
[0020] In the description of the embodiments of the present application, it should be noted that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, which are only for the convenience of describing the embodiments of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the embodiments of the present application. In addition, the terms "first", "second", and "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.
[0021] In the description of the embodiments of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.
[0022] In the embodiments of the present application, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.
[0023] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiments of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.
[0024] The present application proposes an electronic information identification method, device, electronic device, storage medium and computer program product.
[0025] Figure 1 is a flow chart of the electronic information identification method provided in the embodiment of the present application, such as Figure 1 As shown, the electronic information identification method includes: Step 110, performing word segmentation processing on the electronic information to be recognized to obtain at least one word.
[0026] Step 120, input each word into the abnormal information recognition model to obtain the abnormal probability information of each word output by the abnormal information recognition model causing the abnormality of the electronic information to be identified; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula.
[0027] Step 130: determining an abnormality identification result of the electronic information to be identified according to each abnormality probability information.
[0028] It should be noted that the execution subject of the electronic information identification method provided in the embodiment of the present application may be a server, a computer device, etc. The computer device may be, for example, a mobile phone, a tablet computer, a laptop computer, a PDA, an in-vehicle electronic device, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc. It should be noted that the data required to be obtained in this application are all obtained through formal channels after being authorized by the relevant users.
[0029] An electronic information identification device may be provided or connected to the server or computer device of the present application, thereby controlling the electronic information identification device to execute the electronic information identification method of the present application.
[0030] The present application may construct a model based on the Bayesian formula in advance, wherein the Bayesian formula may be specifically shown as follows: , i=1,2,...,n; Among them, P(A i |B) represents the probability that the ith event in an event set A containing n events leads to the occurrence of event B, P(B|A i ) means in A i The probability of B occurring when A occurs is P(A i ) means A i The probability of occurrence, P(B) represents the probability of B occurring, P(A i B) means A i The probability of A occurring simultaneously with B.
[0031] Based on the constructed preset model, this application can collect a large amount of SMS data and email data, including normal SMS, abnormal SMS, normal emails, and abnormal emails, obtain sample electronic information, and label it to obtain information labels for the sample electronic information. The training data set is composed of the sample electronic information and its information labels. Among them, the training data set can be divided into a training set and a test set.
[0032] The labeling method can be binary classification, that is, normal is labeled as 0, abnormal is labeled as 1, and different abnormal types can be labeled for abnormal situations. The abnormal types can be set according to actual rules.
[0033] Clean the text of the sample electronic information, remove irrelevant information such as special characters and punctuation marks, and then perform word segmentation to convert the text into a vocabulary sequence. At the same time, operations such as lemmatization and stemming can also be performed on the vocabulary to reduce the variant forms of the vocabulary.
[0034] Calculate the prior probability of abnormal text in the training set: Count the proportion of the number of abnormal text in the total number of text, denoted as P(abnormal). For example, among 1000 SMS, 200 are abnormal SMS, then P(abnormal) = 200 / 1000 = 0.2.
[0035] Furthermore, count the frequency of each vocabulary in all texts, denoted as P(vocabulary). For example, the vocabulary "deal cards" appears 50 times in 1000 SMS, then P("deal cards") = 50 / 1000 = 0.05.
[0036] Furthermore, calculate the conditional probability of the vocabulary in the abnormal text of the training set: For each vocabulary, count the number of times it appears in the abnormal text, denoted as n(vocabulary|abnormal), and then calculate the conditional probability of the vocabulary appearing in the abnormal text P(vocabulary|abnormal) = n(vocabulary|abnormal) / total number of abnormal text. For example, the vocabulary "deal cards" appears 40 times in 200 abnormal texts, then P("deal cards"|abnormal) = 40 / 200 = 0.2.
[0037] Furthermore, calculate the conditional probability of the vocabulary in the normal text: Similarly, count the number of times the vocabulary appears in the normal text, denoted as n(vocabulary|normal), and calculate its conditional probability P(vocabulary|normal) = n(vocabulary|normal) / total number of normal SMS. Assume that the vocabulary "deal cards" appears 10 times in 800 normal SMS, then P("deal cards"|normal) = 10 / 800 = 0.0125.
[0038] Furthermore, the evidence probability P(vocabulary) is calculated according to the total probability formula, that is, P(vocabulary)=P(vocabulary|abnormal)*P(abnormal)+P(vocabulary|normal)*P(normal). Taking the word "dealing cards" as an example, P(normal)=1-P(abnormal)=0.8, then P("dealing cards")=0.2*0.2+0.8*0.0125=0.05.
[0039] Furthermore, according to the Bayesian formula P(abnormal|vocabulary)=[P(vocabulary|abnormal)*P(abnormal)] / P(vocabulary), the posterior probability that each word causes the text message to be an abnormal text message is calculated. For the word "deal", P(abnormal|"deal")=(0.2*0.2) / 0.05=0.8.
[0040] The prediction results of the model are evaluated through the test set, and evaluation indicators such as accuracy and recall are calculated. According to the evaluation results, the model is optimized, such as adjusting the feature representation of vocabulary, improving the probability calculation method, adding more training data, etc., to improve the prediction performance of the model.
[0041] Thus, an abnormal information recognition model can be obtained by inputting a word to obtain abnormal probability information of the overall information caused by the word. The abnormal information recognition model can determine the probability of different types of abnormalities caused by the word in the overall information.
[0042] Therefore, for the input electronic information to be identified, it is segmented to obtain a vocabulary sequence including at least one vocabulary. Further, for each vocabulary, it is input into the abnormal information identification model, and the abnormal information identification model evaluates the contribution of the vocabulary to the electronic information to be identified as abnormal electronic information based on the posterior probability P (abnormal|vocabulary) calculated previously, and obtains the abnormal probability information of each vocabulary output by the abnormal information identification model that causes the electronic information to be identified to be abnormal. That is, the abnormal probability information includes at least one abnormal type and the abnormal probability corresponding to the corresponding abnormal type.
[0043] After obtaining the abnormal probability information corresponding to each word, the total probability formula can be used in combination with the abnormal probability in the abnormal probability information to determine the total probability that each word causes the electronic information to be identified to be abnormal, and whether the electronic information to be identified is abnormal and the type of abnormality can be determined based on the total probability.
[0044] Specifically, the total probability of each type of anomaly that may be caused by all words can be determined respectively, and then each total probability is compared with a preset probability threshold to determine whether the electronic information to be identified is abnormal, and the abnormal identification result of the abnormal type when an abnormality exists. The preset probability threshold is obtained by adjusting the preset initial probability threshold, and the preset initial probability threshold can be a probability value set according to experience or other methods.
[0045] According to the electronic information identification method of the embodiment of the present application, a preset model is constructed in advance through the Bayesian formula, and the preset model is trained through sample electronic information and its information labels to obtain an abnormal information identification model, so that after the electronic information to be identified is segmented to obtain at least one word, each word is input into the abnormal information identification model, and the abnormal probability information of each word output by the abnormal information identification model causing the electronic information to be identified to be abnormal can be obtained quickly and accurately, and then according to each abnormal probability information, the abnormal identification result of the electronic information to be identified can be quickly and accurately determined, thereby improving the recognition efficiency of electronic information.
[0046] Based on the above embodiment, the electronic information to be identified is segmented to obtain at least one vocabulary, including: The electronic information to be identified is input into a word segmentation model to obtain at least one word output by the word segmentation model; wherein the word segmentation model is trained based on sample electronic information and its word segmentation labels.
[0047] Specifically, when performing word segmentation processing on the electronic information to be identified, the present application can select appropriate word segmentation tools and techniques.
[0048] For example, you can choose a single word segmentation technology, or you can combine multiple word segmentation technologies.
[0049] For word segmentation technology, you can use traditional word segmentation technology or new word segmentation technology.
[0050] In some embodiments, the present application can perform model training through pre-acquired sample electronic information and segmentation labels obtained by annotating the sample electronic information, and obtain a segmentation model that can segment input text after the training is completed. The segmentation label of the sample electronic information may include a vocabulary sequence obtained by vocabulary division of the sample electronic information.
[0051] Since different people may use different vocabulary combinations and expressions when expressing the same or similar meanings, the present application can unify these different forms of expression to the vocabulary level through accurate word segmentation, so that the model can learn the commonalities and rules between words without being restricted by specific text expressions, so that it is convenient for the subsequent rapid and accurate acquisition of the abnormal probability information of each word output by the abnormal information recognition model that causes the abnormality of the electronic information to be identified, and then the abnormal recognition result of the electronic information to be identified can be quickly and accurately determined based on each abnormal probability information, thereby improving the recognition efficiency of electronic information.
[0052] Based on the above embodiment, determining the abnormality identification result of the electronic information to be identified according to each abnormality probability information includes: Adding the abnormality probabilities of the same abnormality type of each word to obtain at least one summed probability; Compare each summed probability with a preset probability threshold to obtain a comparison result; The abnormal identification result of the electronic information to be identified is determined based on the comparison result.
[0053] It should be noted that the abnormality probability information of the present application may include one or more abnormality types and their abnormality probabilities. For example, the abnormality probability information of vocabulary M may include the abnormality probability of abnormality type X, the abnormality probability of abnormality type Y, and the abnormality probability of abnormality type Z, etc.; the abnormality probability information of vocabulary N may include the abnormality probability of abnormality type X, the abnormality probability of abnormality type Y, and the abnormality probability of abnormality type Z, etc.
[0054] Therefore, for each word, the abnormality probabilities of the same abnormality type can be summed to obtain a corresponding number of summed probabilities. Based on the above example, the abnormality probabilities of abnormality type X of each word can be summed to obtain the summed probability of abnormality type X; the abnormality probabilities of abnormality type Y of each word can be summed to obtain the summed probability of abnormality type Y; the abnormality probabilities of abnormality type Z of each word can be summed to obtain the summed probability of abnormality type Z.
[0055] Furthermore, each summed probability is compared with a preset probability threshold, and a comparison result including a magnitude relationship between each summed probability and the preset probability threshold is obtained.
[0056] The present application can obtain event probability samples through defined events, first define a higher event trigger probability as a preset initial probability threshold, and use this as a target to run a batch of undefined event samples, and filter the data in the undefined event samples to define the trigger probability of events in some samples, and then gradually reduce the event trigger probability, and continuously perform a cyclic screening of undefined events until samples that do not belong to the event are screened out, and then gradually refine the trigger probability, and finally obtain the final trigger probability as the preset probability threshold.
[0057] In some embodiments, for example, a higher event trigger probability can be defined as an initial probability threshold, and the setting of this threshold is usually based on experience or a preliminary understanding of the problem. For example, in the identification of spam text messages, if it is initially believed that certain text messages that clearly contain specific sensitive words or typical fraud expressions have a higher probability of being abnormal, the trigger probability corresponding to these words or expressions can be set higher, such as 0.8 or 0.9.
[0058] Furthermore, this higher probability threshold is used as a target to screen undefined event samples. When the trigger probability of the words or phrases contained in the sample is higher than this threshold, the sample is preliminarily determined to belong to the event (i.e., abnormal event), thereby defining a part of samples that are clearly abnormal.
[0059] Furthermore, after the initial screening, the event trigger probability is gradually reduced, such as by 0.1 or 0.05 each time, and then the remaining undefined event samples are screened again with the new threshold. As the threshold is lowered, more samples will be included in the category of possible abnormal events.
[0060] By repeatedly cycling through this process of lowering the threshold and screening, the set of defined abnormal samples is continuously expanded, and at the same time, some samples that are in a marginal state are gradually identified, that is, they may be judged as abnormal or normal under different thresholds.
[0061] In the process of continuously lowering the threshold, when a certain threshold is reached, samples that clearly do not meet the characteristics of abnormal events will begin to appear in the screened samples, and these samples should not be judged as abnormal under normal circumstances. For example, some text messages composed of ordinary daily language may be misjudged as abnormal under too low a threshold, but in fact they are not abnormal events.
[0062] After those samples that are obviously not abnormal events are identified, the threshold becomes an important reference point, which marks the approximate boundary between abnormal events and normal events.
[0063] After determining the preliminary boundary, make more refined adjustments within the threshold range near this boundary, such as changing the threshold in smaller steps (such as 0.01 or 0.02) and re-screening and evaluating the samples.
[0064] The final probability threshold is determined by observing the various evaluation indicators of the model under different refinement thresholds, such as accuracy, recall, F1 value, etc. In general, a threshold that achieves a good balance between accuracy and recall is selected as the probability threshold for the final judgment of whether the text is abnormal. For example, when the threshold is set to 0.3, the accuracy is 0.9, the recall is 0.7, and the F1 value is 0.8. When the threshold is adjusted to 0.35, the accuracy increases to 0.92, but the recall is reduced to 0.65, and the F1 value becomes 0.76. At this time, it can be considered that 0.3 is a relatively better final probability threshold based on comprehensive considerations.
[0065] Among them, the F1 value is a comprehensive evaluation indicator used to measure the performance of the classification model. It comprehensively considers the accuracy and recall rate of the model and can more comprehensively evaluate the performance of the model in the classification task.
[0066] Furthermore, it is possible to determine whether the electronic information to be identified has an abnormality based on the size relationships in the comparison result, and the abnormality identification result of the abnormality type when an abnormality exists.
[0067] Further, determining the abnormal identification result of the electronic information to be identified according to the comparison result includes: If any sum probability in the comparison result is greater than the preset probability threshold, the electronic information to be identified is determined to be abnormal information, and the abnormal type corresponding to the sum probability greater than the preset probability threshold is determined as the abnormal type of the abnormal information.
[0068] After obtaining the comparison result, if it is determined that there is any sum probability in the comparison result that is greater than the preset probability threshold, the electronic information to be identified can be determined as abnormal information. After determining that the electronic information to be identified is abnormal information, the abnormal type corresponding to the sum probability greater than the preset probability threshold is determined as the abnormal type of the abnormal information.
[0069] For example, if the summed probability of the abnormal type Y in each word is greater than a preset probability threshold, the electronic information to be identified can be determined as abnormal information. Further, the abnormal type of the electronic information to be identified can be determined. Specifically, the abnormal type corresponding to all the summed probabilities greater than the preset probability threshold can be determined, and the corresponding abnormal type can be determined as the abnormal type of the abnormal information.
[0070] The present application can quickly and accurately determine the abnormal identification result of the electronic information to be identified based on the comparison between the abnormal probability in each abnormal probability information and the preset probability threshold, thereby improving the identification efficiency of the electronic information.
[0071] Based on the above embodiment, after determining the abnormality identification result of the electronic information to be identified according to each abnormality probability information, the method further includes: Determine a solution for handling the electronic information to be identified based on the abnormal identification result of the electronic information to be identified.
[0072] After obtaining the abnormal identification result of the electronic information to be identified, the present application can formulate different processing schemes for abnormal information and normal information. Among them, for normal information, a non-interference processing scheme can be formulated.
[0073] Abnormal information (such as spam text messages or spam emails) can be processed, which may include filtering and intercepting, marking and reminding.
[0074] Specifically, devices such as mail servers or SMS gateways can be used to automatically filter out abnormal junk SMS or junk emails based on preset rules and algorithms, preventing them from entering the user's inbox or SMS inbox. This method can effectively prevent users from being disturbed by a large amount of junk information and improve the efficiency of users in obtaining valid information.
[0075] The source addresses, numbers or domain names of spam text messages or spam emails can be added to the blacklist, and any messages from these blacklists will be directly blocked without the need for abnormal detection again. Users and network service providers can jointly maintain and update the blacklist to enhance the blocking effect.
[0076] For spam text messages or spam emails that have entered the inbox but are identified as abnormal, you can instruct the email client or SMS application to mark them specially, such as adding a specific logo to the email subject or the beginning of the SMS, to remind users that the information may be spam and to handle it with caution.
[0077] Users can be reminded through pop-up windows, notification bar messages, etc. that they have received spam text messages or spam emails, so that they can understand them in time and take corresponding measures, such as directly deleting or reporting them.
[0078] This application determines a processing plan for the electronic information to be identified based on the abnormal identification results of the electronic information to be identified, which can reduce the risk of users encountering fraud. By preventing the spread of these malicious information at the source, it can also enhance the security of user devices and information.
[0079] The electronic information identification device provided in the present application is described below. The electronic information identification device described below and the electronic information identification method described above can be referenced to each other.
[0080] Furthermore, the present application also provides an electronic information identification device.
[0081] The electronic information identification device comprises: A word segmentation module, used for performing word segmentation processing on the electronic information to be identified to obtain at least one word; The recognition module is used to input each word into the abnormal information recognition model to obtain the abnormal probability information that each word output by the abnormal information recognition model causes the abnormality of the electronic information to be recognized; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula; The determination module is used to determine the abnormality identification result of the electronic information to be identified according to each abnormality probability information.
[0082] The electronic information identification device of the present application constructs a preset model through the Bayesian formula in advance, and obtains an abnormal information identification model by training the preset model through sample electronic information and its information labels, so that after the electronic information to be identified is segmented to obtain at least one word, each word is input into the abnormal information identification model, and the abnormal probability information of each word output by the abnormal information identification model causing the electronic information to be identified to be abnormal can be obtained quickly and accurately, and then the abnormal identification result of the electronic information to be identified can be quickly and accurately determined according to each abnormal probability information, thereby improving the recognition efficiency of electronic information.
[0083] In one embodiment, the word segmentation module is specifically used for: The electronic information to be identified is input into a word segmentation model to obtain at least one word output by the word segmentation model; wherein the word segmentation model is trained based on sample electronic information and its word segmentation labels.
[0084] In one embodiment, the determining module is specifically used to: Adding the abnormality probabilities of the same abnormality type of each word to obtain at least one summed probability; Compare each summed probability with a preset probability threshold to obtain a comparison result; The abnormality identification result of the electronic information to be identified is determined according to the comparison result.
[0085] In one embodiment, the determining module includes a determining unit, and the determining unit is used to: If any sum probability in the comparison result is greater than the preset probability threshold, the electronic information to be identified is determined to be abnormal information, and the abnormal type corresponding to the sum probability greater than the preset probability threshold is determined as the abnormal type of the abnormal information.
[0086] In one embodiment, the determining module is further configured to: A processing scheme for the electronic information to be identified is determined according to the abnormal identification result of the electronic information to be identified.
[0087] Figure 2 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 2As shown, the electronic device may include: a processor 210, a communication interface 220, a memory 230 and a communication bus 240, wherein the processor 210, the communication interface 220 and the memory 230 communicate with each other via the communication bus 240. The processor 210 may call the logic instructions in the memory 230 to execute the following method: performing word segmentation processing on the electronic information to be recognized to obtain at least one vocabulary; Input each word into the abnormal information recognition model to obtain abnormal probability information of each word output by the abnormal information recognition model causing the abnormality of the electronic information to be recognized; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula; The abnormality identification result of the electronic information to be identified is determined according to each abnormality probability information.
[0088] In addition, the logic instructions in the above-mentioned memory 230 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the relevant technology. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.
[0089] In another aspect, the embodiments of the present application further provide a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method provided by the above embodiments is implemented, for example, including: performing word segmentation processing on the electronic information to be identified to obtain at least one vocabulary; Input each word into the abnormal information recognition model to obtain abnormal probability information of each word output by the abnormal information recognition model causing the abnormality of the electronic information to be recognized; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula; The abnormality identification result of the electronic information to be identified is determined according to each abnormality probability information.
[0090] In another aspect, the embodiments of the present application further provide a computer program product having a computer program stored thereon, and when the computer program is executed by a processor, the method provided by the above embodiments is implemented, for example, including: performing word segmentation processing on the electronic information to be identified to obtain at least one vocabulary; Input each word into the abnormal information recognition model to obtain abnormal probability information of each word output by the abnormal information recognition model causing the abnormality of the electronic information to be recognized; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula; The abnormality identification result of the electronic information to be identified is determined according to each abnormality probability information.
[0091] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0092] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.
[0093] Finally, it should be noted that the above implementation modes are only used to illustrate the present application, rather than to limit the present application. Although the present application is described in detail with reference to the embodiments, a person skilled in the art should understand that various combinations, modifications or equivalent substitutions of the technical solutions of the present application do not depart from the spirit and scope of the technical solutions of the present application.
Claims
1. A method for identifying electronic information, characterized in that: include: Performing word segmentation processing on the electronic information to be identified to obtain at least one word; Input each word into the abnormal information recognition model to obtain abnormal probability information of each word output by the abnormal information recognition model causing the abnormality of the electronic information to be recognized; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula; The abnormality identification result of the electronic information to be identified is determined according to each abnormality probability information.
2. The electronic information identification method according to claim 1, characterized in that: The abnormal probability information includes at least one abnormal type and an abnormal probability corresponding to the corresponding abnormal type; the abnormal identification result of the electronic information to be identified is determined according to each abnormal probability information, including: Adding the abnormality probabilities of the same abnormality type of each word to obtain at least one summed probability; Compare each summed probability with a preset probability threshold to obtain a comparison result; The abnormality identification result of the electronic information to be identified is determined according to the comparison result.
3. The electronic information identification method according to claim 2, characterized in that: Determining the abnormality recognition result of the electronic information to be recognized according to the comparison result includes: If any sum probability in the comparison result is greater than the preset probability threshold, the electronic information to be identified is determined to be abnormal information, and the abnormal type corresponding to the sum probability greater than the preset probability threshold is determined as the abnormal type of the abnormal information.
4. The electronic information identification method according to claim 3, characterized in that: The preset probability threshold is adjusted based on a preset initial probability threshold.
5. The electronic information identification method according to claim 1, characterized in that: The electronic information to be identified is segmented to obtain at least one word, including: The electronic information to be identified is input into a word segmentation model to obtain at least one word output by the word segmentation model; wherein the word segmentation model is trained based on sample electronic information and its word segmentation labels.
6. The electronic information identification method according to claim 1, characterized in that: After determining the abnormality identification result of the electronic information to be identified according to each abnormality probability information, the method further includes: A processing scheme for the electronic information to be identified is determined according to the abnormal identification result of the electronic information to be identified.
7. An electronic information identification device, characterized in that: include: A word segmentation module, used for performing word segmentation processing on the electronic information to be identified to obtain at least one word; The recognition module is used to input each word into the abnormal information recognition model to obtain the abnormal probability information that each word output by the abnormal information recognition model causes the abnormality of the electronic information to be recognized; wherein the abnormal information recognition model is obtained by training a preset model through sample electronic information and its information labels; the preset model is constructed based on the Bayesian formula; The determination module is used to determine the abnormality identification result of the electronic information to be identified according to each abnormality probability information.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the electronic information identification method according to any one of claims 1 to 6 is implemented.
9. A storage medium, the storage medium being a non-transitory computer-readable storage medium, on which a computer program is stored, characterized in that: When the computer program is executed by a processor, the electronic information identification method as described in any one of claims 1 to 6 is implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the electronic information identification method according to any one of claims 1 to 6 is implemented.