Remote communication-based on-hook message adaptive pushing system

Abstract

The invention relates to the technical field of communication message pushing, in particular to an on-hook message self-adaptive pushing system based on remote communication, and the system comprises a communication data processing module which is used for carrying out the character recognition and word segmentation of information in each remote communication process; obtaining the word polarity of each segmented word appearing each time; the communication feature extraction module is used for determining the information tendency quantity and the word segmentation tendency change rate of each word segmentation in each remote communication process; obtaining the word segmentation tense interference rate of each word segmentation in each remote communication process; determining the trend variation of each segmented word in the current remote communication process; the on-hook message pushing module is used for obtaining the weight coefficient of each segmented word in the current remote communication process; correcting the TF value in the TF-IDF algorithm, and obtaining the TF-IDF value of each segmented word in the current remote communication process; and screening each segmented word through the TF-IDF value, matching the segmented words with recommendation information, and pushing the on-hook message. According to the invention, the on-hook message pushing accuracy is improved.

Description

Technical Field

[0001] The present application relates to the technical field of communication message push technology, and in particular to an on-hook message adaptive push system based on remote communication. Background Art

[0002] Idle messages usually refer to messages that are automatically sent when a user leaves or is no longer active in some online services or platforms. Idle message push is designed to improve user experience, increase user engagement and improve service efficiency.

[0003] Existing push notifications often lack an understanding of user needs and are unable to accurately grasp their true needs. Push algorithms overly rely on historical data and fail to promptly consider the user's current context, resulting in push notifications that may be irrelevant to the user's current interests and needs. Furthermore, the TF-IDF algorithm relies solely on word frequency when calculating word importance, making the extraction of user-interested information in remote communications less reliable, leading to low accuracy in push notifications. Summary of the Invention

[0004] In order to solve the above technical problems, the present application provides an on-hook message adaptive push system based on remote communication to solve the existing problems.

[0005] The present application discloses a remote communication-based adaptive push system for on-hook messages using the following technical solutions:

[0006] One embodiment of the present application provides a system for adaptively pushing on-hook messages based on remote communication, the system comprising:

[0007] The communication data processing module is used to perform text recognition and word segmentation on the information in each remote communication process; obtain the word polarity of each word at each occurrence based on the emotional tendency of each word at each occurrence;

[0008] A communication feature extraction module is used to determine the information tendency of each segmentation during each remote communication by combining the word polarity of all occurrences of each segmentation during each remote communication; analyze the difference in the information tendency of each segmentation during each remote communication relative to the first remote communication, and determine the segmentation tendency change rate of each segmentation during each remote communication;

[0009] Using the degree of temporal importance, weights are added to the segmentation tendency change rates to obtain the segmentation temporal interference rate of each segmentation during each telecommunication process; the changing trend of the segmentation temporal interference rate of each segmentation during all previous telecommunication processes is analyzed to determine the trend change amount of each segmentation during the current telecommunication process;

[0010] The hang-up message push module obtains a weight coefficient for each word in the current remote communication process based on the distribution of the information tendency of each word in the historical remote communication process and the trend change; modifies the TF value in the TF-IDF algorithm using the weight coefficient, and uses the modified TF-IDF algorithm to obtain the TF-IDF value of each word in the current remote communication process;

[0011] Each word segment is filtered by the TF-IDF value, matched with the recommended information, and the hang-up message is pushed.

[0012] In one embodiment, obtaining the word polarity of each word segment at each occurrence includes:

[0013] The semantic analysis algorithm is used to obtain the sentiment classification results of each segmentation word each time it appears. The word polarity of the segmentation word belonging to positive vocabulary is set to 1, the word polarity of the segmentation word belonging to negative vocabulary is set to -1, and the word polarity of the segmentation word belonging to neither positive vocabulary nor negative vocabulary is set to 0.

[0014] In one embodiment, determining the information tendency includes:

[0015] For each remote communication, the sum of the word polarities of all occurrences of each participle is calculated, and the information tendency is positively correlated with the sum and the number of occurrences of each participle.

[0016] In one embodiment, the word segmentation tendency change rate is positively correlated with the difference of each word segmentation, and negatively correlated with the information tendency amount of each word segmentation during the first remote communication.

[0017] In one embodiment, determining the word segmentation tense interference rate includes:

[0018] The cumulative sum of the sequence numbers of the current and all previous remote communication processes is calculated, and the ratio of the sequence number of each remote communication process to the cumulative sum is calculated. The word segmentation tense interference rate is the product of the word segmentation tendency change rate and the ratio.

[0019] In one embodiment, determining the trend change amount includes:

[0020] Obtaining the slope of the fitted straight line of the word segmentation tense interference rate of any word segmentation in the current and all previous remote communication processes;

[0021] Calculate the symbolized result of the sum of the temporal interference rates of any word segmentation in the current and all previous remote communication processes, and the trend change of any word segmentation in the current remote communication process is the product of the slope and the symbolized result.

[0022] In one embodiment, determining the weight coefficient includes:

[0023] The normalized result of the mean value of the information tendency of each word in the current and all previous remote communication processes in the current remote communication process is calculated, and the weight coefficient is the product of the normalized result and the normalized result of the trend change amount.

[0024] In one embodiment, the TF value in the TF-IDF algorithm is corrected by the weight coefficient, and the calculation method is:

[0025] Where tf(t) is the TF value of the tth word segmentation after correction in the current remote communication process, W t is the weight coefficient of the t-th segmentation in the current remote communication process, f(t) is the frequency of occurrence of the t-th segmentation in the current remote communication process, and m is the number of segmentations in the current remote communication process.

[0026] In one embodiment, filtering each word by the TF-IDF value includes:

[0027] Obtaining each segmentation that does not appear in the current telecommunication process but appears in the historical telecommunication process as a historical segmentation, and multiplying the TF-IDF value of each historical segmentation by a preset attenuation coefficient as the relative importance of each historical segmentation;

[0028] The TF-IDF values of all the segmented words in the current remote communication process and the relative importance of all the historical segmented words are arranged in descending order, and a preset number of segmented words that are ranked at the top are used as the segmented words obtained by screening.

[0029] In one embodiment, matching the recommendation information and pushing the hang-up message includes:

[0030] The similarity between the filtered segmented words and all segmented words of each on-hook message in the on-hook message push library is calculated, and the on-hook message with the highest similarity is pushed.

[0031] This application has at least the following beneficial effects:

[0032] The present application performs text recognition and word segmentation on the information in each remote communication process; it can accurately extract key information from the communication content, laying the foundation for subsequent sentiment analysis and feature extraction, and obtain the word polarity of each segmentation each time it appears based on the sentiment tendency of each segmentation each time it appears; by analyzing the sentiment tendency of each segmentation, the sentiment polarity of each word can be quantified; then, combined with the word polarity of all occurrences of each segmentation in each remote communication process, the information tendency of each segmentation in each remote communication process is determined; the determination of the information tendency helps to understand the overall interest change trend of the communication content and improve the accuracy of key information extraction in the remote communication process; the difference in the information tendency of each segmentation in each remote communication process relative to the first remote communication is analyzed to determine the segmentation tendency change rate of each segmentation in each remote communication process; the segmentation tendency change rate dynamically reflects the diachronic evolution of the emotional expression of words by comparing the information tendency difference between the current and first communications; using the degree of importance in time, the segmentation tendency change rate is weighted to obtain the segmentation temporal interference rate of each segmentation in each remote communication process; the segmentation temporal interference rate strengthens the timeliness influence of recent communication data and weakens the interference of historical noise, so that trend analysis The method is more in line with actual scenario changes; the changing trend of the temporal interference rate of each segmented word in all previous remote communication processes is analyzed to determine the trend change amount of each segmented word in the current remote communication process; the trend change amount identifies the potential evolution direction of the user's communication interest and provides a basis for dynamically adjusting the recommendation strategy; based on the distribution of the information tendency amount of each segmented word in the current remote communication process and the trend change amount, a weight coefficient of each segmented word in the current remote communication process is obtained; the TF value in the TF-IDF algorithm is corrected by the weight coefficient, and the TF-IDF value of each segmented word in the current remote communication process is obtained using the corrected TF-IDF algorithm. By integrating the historical tendency distribution and the real-time trend amount, the TF value simultaneously reflects the global importance and local sentiment mutation characteristics of the word, improves the discrimination of key words, and solves the defect of the traditional TF-IDF algorithm that only relies on word frequency and ignores semantic tendency; each segmented word is screened by the TF-IDF value and matched with recommended information to push the hang-up message, emphasizing the importance of the user's current communication content to the hang-up message push, reducing the interference of invalid information, and improving the accuracy and applicability of the hang-up message push. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0034] Figure 1 A block diagram of a remote communication-based adaptive push system for on-hook messages provided by this application;

[0035] Figure 2 This is the block diagram of the communication feature extraction module;

[0036] Figure 3 This is the block diagram of the hang-up message push module. DETAILED DESCRIPTION

[0037] To further illustrate the technical means and effectiveness of this application to achieve the intended invention objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of an adaptive push system for on-hook messages based on remote communication proposed in this application. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0038] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0039] The following describes in detail a specific solution of an on-hook message adaptive push system based on remote communication provided by the present application with reference to the accompanying drawings.

[0040] An embodiment of the present application provides an on-hook message adaptive push system based on remote communication. Specifically, the following on-hook message adaptive push system based on remote communication is provided. Figure 1 The system includes: a communication data processing module 101, a communication feature extraction module 102, and an on-hook message pushing module 103.

[0041] The communication data processing module 101 is used to perform text recognition and word segmentation on the information in each remote communication process; and obtain the word polarity of each word at each occurrence based on the emotional tendency of each word at each occurrence.

[0042] This embodiment obtains the text data of each user's remote communication process by using the remote communication background database, and uses the jieba word segmentation tool to segment the text in each remote communication process to obtain each word segmentation of the user's remote communication process.

[0043] It should be noted that if the user uses voice signals during remote communication, the voice signals are converted into text through voice recognition. Among them, the use of the remote communication background database to obtain the text data of the user during each remote communication process, and voice recognition are both existing public technologies and will not be described in detail in this embodiment.

[0044] For each segmented word in each remote communication process of the user, the emotional tendency of each segmented word is analyzed each time it appears to obtain the word polarity of each segmented word each time it appears. Specifically, this embodiment adopts a hybrid model based on the Transformer encoder and the Long Short-Term Memory (LSTM) network to perform sentiment classification on each segmented word, setting the word polarity of the segmented words belonging to positive vocabulary to 1, setting the word polarity of the segmented words belonging to negative vocabulary to -1, and setting the word polarity of the segmented words that are neither positive nor negative vocabulary to 0. Among them, the hybrid model based on the Transformer encoder and the Long Short-Term Memory (LSTM) network is an existing public technology, and the specific process will not be repeated. The implementer can choose other existing feasible sentiment classification models at his own discretion, and this embodiment does not limit it here.

[0045] The communication feature extraction module 102 (1) is used to determine the information tendency of each segmentation during each remote communication by combining the word polarity of all occurrences of each segmentation during each remote communication; analyze the difference in the information tendency of each segmentation during each remote communication relative to the first remote communication, and determine the segmentation tendency change rate of each segmentation during each remote communication.

[0046] It should be understood that during a remote communication, the more interested a user is in the information and the more times the information is repeated, the more likely the segmented words in the text information are to be positive after sentiment analysis. Therefore, when recommending hang-up messages to users, priority should be given to those messages that are determined to be positive and have a high repetition frequency after sentiment analysis. This not only accurately meets user needs but also improves user experience. Therefore, this embodiment calculates the information tendency of each segmented word in each remote communication process based on the word polarity of all occurrences of each segmented word. The specific calculation method is:

[0047] Where, Hq i Cs is the information tendency of the i-th word in each remote communication process, i is the number of occurrences of the i-th word in each remote communication process, Yu i,j is the word polarity of the jth occurrence of the i-th segment in each remote communication process.

[0048] In information analysis of user communications, high-frequency segmented words are more effective in reflecting user interests. The higher the frequency of a segmented word, the higher the attention paid to the information. This means that the information tendency calculated from these high-frequency segmented words has a higher reference value. When analyzing information tendency of user behavior, the positive tendency of high-frequency segmented words can effectively improve the accuracy of message recommendations.

[0049] This embodiment obtains each segmentation in each remote communication process of the user, calculates the information tendency of each segmentation in each remote communication process of the user, obtains each segmentation in the current remote communication process and all previous remote communication processes, takes the i-th segmentation as an example, if the i-th segmentation does not appear in the remote communication process, then the information tendency of the i-th segmentation in the remote communication process is set to 0, thus obtaining the information tendency of the i-th segmentation in all remote communication processes, and analyzing the change characteristics of the information tendency of the i-th segmentation over time. If the information tendency of the segmentation continues to increase, and the faster the increase rate, it means that the user is more interested in the matching information of the segmentation.

[0050] Based on the above analysis, this embodiment calculates the segmentation tendency change rate of each segmentation word in each remote communication process. The specific calculation method is:

[0051] Where Zt i,k Po is the change rate of the i-th word segmentation tendency in the k-th remote communication process, i,k is the information tendency of the kth telecommunicating process of the i-th segmentation, Po i,1 is the information tendency of the first remote communication process of the i-th word segment, τ is a preset value greater than 0 to avoid the denominator being 0. In this embodiment, τ=0.01. The implementer can set it according to the actual situation, and this embodiment does not impose any restrictions on this.

[0052] It should be understood that during user telecommunications, the rate of change in information propensity can effectively reflect the changing trends in user interests. When the information propensity of a segmented word increases over time, and the faster the rate of increase, the greater the user's interest in the idle messages associated with that segmented word. This means that message recommendations based on these high-incremental-rate segmented words have a higher reference value. When analyzing user communication data for interests, high-incremental-rate propensity changes can effectively improve the accuracy of recommendations, thereby better meeting the user's personalized idle message recommendations.

[0053] (2) Using the degree of importance in time, weights are added to the segmentation tendency change rate to obtain the segmentation temporal interference rate of each segmentation in each remote communication process; the changing trend of the segmentation temporal interference rate of each segmentation in all previous remote communication processes is analyzed to determine the trend change amount of each segmentation in the current remote communication process.

[0054] However, user preferences and interests may change dynamically over time. Therefore, information closer to the current telecommuting process is more important for assessing user interests than historical telecommuting information. Time sensitivity is particularly important in precision recommendations because user interests can change significantly over time. For example, user behavior information in the short term may better reflect the user's immediate interests, while historical data may contain outdated preferences. Therefore, recommendation information needs to rely more on recent data to capture the changing trends of user interests.

[0055] Based on the above analysis, this embodiment calculates the word segmentation temporal interference rate of each word segmentation in each remote communication process. The specific calculation process is:

[0056] Where Zy i,k is the temporal interference rate of the i-th word segmentation in the k-th remote communication process, k is the sequence number of the remote communication process, Zt i,k is the change rate of the segmentation tendency of the i-th segmentation word in the k-th remote communication process, k n is the sequence number of the nth remote communication process, and N is the number of the current and all previous remote communication processes.

[0057] It should be noted that the sequence number of the remote communication process is the time sequence of the remote communication process. For example, the sequence number of the first remote communication process is 1, the sequence number of the second remote communication process is 2, the sequence number of the third remote communication process is 3, and so on.

[0058] It should be understood that this embodiment assigns a time weight to each word segmentation tendency change rate to emphasize information close to the current moment, which can more accurately capture the user's recent interests and improve the accuracy of idle message recommendations.

[0059] In addition, in order to measure the changing trend of the temporal interference rate of each segmentation word in all previous remote communication processes, this embodiment uses the least squares method to obtain the fitting straight line of the temporal interference rate of each segmentation word in all previous remote communication processes in the time sequence, and combines the numerical distribution of the temporal interference rate of each segmentation word in all remote communication processes to calculate the trend change amount of each segmentation word in the current remote communication process. The specific calculation method is:

[0060] Where A i is the trend change of the i-th word in the current remote communication process, Zy i,k is the temporal interference rate of the i-th word segmentation in the k-th remote communication process, N is the number of the current and all previous remote communication processes, a iis the slope of the fitting line of the i-th word segment in the current remote communication process, sign() is a sign function, which takes the value of 1 when the value is positive, -1 when the value is negative, and 0 when the value is 0.

[0061] It should be understood that during user remote communication, the changing trend of the temporal interference rate of each word segmentation over time can effectively reflect the changing direction and intensity of the user's interest. By calculating the trend change of the word segmentation, the dynamic changes of the user's interest can be captured more accurately. When the trend change of the word segmentation is positive, it indicates that the user's interest in the information is increasing; when the trend change of the word segmentation is negative, it indicates that the user's interest in the information is decreasing; and when the trend change of the word segmentation is zero, it indicates that the user's interest is relatively stable. The trend change of the word segmentation can provide more accurate user interest dynamics for message recommendation, thereby optimizing the recommendation effect and better meeting user needs. The block diagram of the communication feature extraction module is shown in the figure. Figure 2 shown.

[0062] The on-hook message push module 103 (1) obtains a weight coefficient of each word in the current remote communication process based on the distribution of the information tendency of each word in the historical remote communication process and the trend change; corrects the TF value in the TF-IDF algorithm by the weight coefficient, and obtains the TF-IDF value of each word in the current remote communication process using the corrected TF-IDF algorithm.

[0063] In order to emphasize the importance of information in the user's current remote communication process and emphasize the consideration of the user's current needs, this embodiment focuses on analyzing the numerical distribution of the information tendency of each segmentation in the current and previous remote communication processes, and calculates the weight coefficient of each segmentation in the current remote communication process in combination with the trend change. The specific calculation method is:

[0064] Where W i is the weight coefficient of the i-th word in the current remote communication process, Po i,k is the information tendency of the kth remote communication process of the i-th segmentation, N is the number of the current and all previous remote communication processes, A i is the trend change of the i-th word in the current remote communication process, and sig() is the sigmoid normalization function.

[0065] It should be understood that when the information tendency of a segmentation word in all remote communication processes is very large, it means that the user is more interested in the recommended information corresponding to the segmentation word. The larger the weight coefficient of the segmentation word is, the higher the importance of the corresponding segmentation word is. When pushing subsequent hang-up messages, more consideration should be given to the relevance of the segmentation word, so that the weight coefficient of the segmentation word in the user's remote communication process is calculated more accurately, and the hang-up message recommended to the user is more accurate.

[0066] Therefore, this embodiment modifies the calculation of the TF value of the TF-IDF algorithm, specifically:

[0067] Where tf(t) is the TF value of the tth word segmentation after correction in the current remote communication process, W t is the weight coefficient of the t-th segmentation in the current remote communication process, f(t) is the frequency of occurrence of the t-th segmentation in the current remote communication process, and m is the number of segmentations in the current remote communication process.

[0068] Then, the TF-IDF algorithm is used to calculate the TF-IDF value of each word in the current long-distance communication process, combining the word frequency weights of the word segmentation in the current long-distance communication process. This value reflects the importance of each word segmentation in the current long-distance communication process. The calculation of the TF-IDF algorithm is a well-known technology, and the specific process is not described in detail here.

[0069] (2) Filter each word by the TF-IDF value, match it with the recommended information, and push the hang-up message.

[0070] Each segmentation word that does not appear in the current remote communication process but appears in the historical remote communication process is obtained and recorded as a historical segmentation word. The product of the TF-IDF value of each historical segmentation word and a preset attenuation coefficient is used as the relative importance of each historical segmentation word. The purpose is to reduce the importance of the segmentation word in the historical remote communication process for the hang-up message recommendation, so that the hang-up message adaptive push system takes more consideration of the segmentation word in the user's current remote communication process.

[0071] It should be noted that the value range of the attenuation coefficient is [0.7, 0.95]. In this embodiment, the value of the attenuation coefficient is 0.9. The implementer can set it within the value range according to actual conditions, and this embodiment does not impose any restrictions here.

[0072] The TF-IDF values of all word segmentations in the current remote communication process and the relative importance of all historical word segmentations are arranged in descending order. At the same time, the hang-up messages in the hang-up message push library are also segmented using the jieba word segmentation tool. All word segmentations of each hang-up message and the word segmentations after the descending order are converted into word vectors, and the similarity between the word vectors of all word segmentations of each hang-up message and the word vectors of the first preset number of word segmentations after the descending order is calculated. Specifically, the first preset number of word segmentations is ensured to be equal to the number of word segmentations of each hang-up message. For example, if there is a hang-up message with 5 word segmentations, then when calculating the similarity with the hang-up message, the first 5 word segmentations after the descending order are selected for calculation.

[0073] It should be noted that in this embodiment, the cosine similarity is used to calculate the similarity between word vectors. The implementer can choose other existing feasible similarity calculation methods, such as the Pearson correlation coefficient.

[0074] Finally, the on-hook message with the highest similarity in the on-hook message push library is pushed to the user, completing the adaptive push of the on-hook message and improving the accuracy of the on-hook message push. Figure 3 shown.

[0075] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0076] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0077] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them. Modifications to the technical solutions described in the aforementioned embodiments, or equivalent replacements of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A remote communication-based adaptive push system for on-hook messages, characterized in that: The system includes: The communication data processing module is used to perform text recognition and word segmentation on the information in each remote communication process; obtain the word polarity of each word each time it appears based on the emotional tendency of each word each time it appears; A communication feature extraction module is used to determine the information tendency of each segmentation during each remote communication by combining the word polarity of all occurrences of each segmentation during each remote communication; analyze the difference in the information tendency of each segmentation during each remote communication relative to the first remote communication, and determine the segmentation tendency change rate of each segmentation during each remote communication; Using the degree of temporal importance, weights are added to the segmentation tendency change rates to obtain the segmentation temporal interference rate of each segmentation during each telecommunication process; the changing trend of the segmentation temporal interference rate of each segmentation during all previous telecommunication processes is analyzed to determine the trend change amount of each segmentation during the current telecommunication process; The hang-up message push module obtains a weight coefficient for each word in the current remote communication process based on the distribution of the information tendency of each word in the historical remote communication process and the trend change; modifies the TF value in the TF-IDF algorithm using the weight coefficient, and uses the modified TF-IDF algorithm to obtain the TF-IDF value of each word in the current remote communication process; Each word segment is filtered by the TF-IDF value, matched with the recommended information, and the hang-up message is pushed.

2. The remote communication-based adaptive push system for on-hook messages according to claim 1, characterized in that: The step of obtaining the word polarity of each word at each occurrence includes: The semantic analysis algorithm is used to obtain the sentiment classification results of each segmentation word each time it appears. The word polarity of the segmentation word belonging to positive vocabulary is set to 1, the word polarity of the segmentation word belonging to negative vocabulary is set to -1, and the word polarity of the segmentation word belonging to neither positive vocabulary nor negative vocabulary is set to 0.

3. The remote communication-based on-hook message adaptive push system according to claim 1, characterized in that: Determining the information tendency includes: For each remote communication, the sum of the word polarities of all occurrences of each participle is calculated, and the information tendency is positively correlated with the sum and the number of occurrences of each participle.

4. The self-adaptive push system for on-hook messages based on remote communication according to claim 1, characterized in that: The segmentation tendency change rate is positively correlated with the difference of each segmentation, and negatively correlated with the information tendency amount of each segmentation during the first remote communication.

5. The remote communication-based on-hook message adaptive push system according to claim 1, characterized in that: The determination of the word segmentation tense interference rate includes: The cumulative sum of the sequence numbers of the current and all previous remote communication processes is calculated, and the ratio of the sequence number of each remote communication process to the cumulative sum is calculated. The word segmentation tense interference rate is the product of the word segmentation tendency change rate and the ratio.

6. The remote communication-based on-hook message adaptive push system according to claim 1, characterized in that: Determining the trend change amount includes: Obtaining the slope of the fitted straight line of the word segmentation tense interference rate of any word segmentation in the current and all previous remote communication processes; Calculate the symbolized result of the sum of the temporal interference rates of any word segmentation in the current and all previous remote communication processes, and the trend change of any word segmentation in the current remote communication process is the product of the slope and the symbolized result.

7. The remote communication-based on-hook message adaptive push system according to claim 1, characterized in that: Determining the weight coefficient includes: The normalized result of the mean value of the information tendency of each word in the current and all previous remote communication processes in the current remote communication process is calculated, and the weight coefficient is the product of the normalized result and the normalized result of the trend change amount.

8. The remote communication-based on-hook message adaptive push system according to claim 1, characterized in that: The TF value in the TF-IDF algorithm is corrected by the weight coefficient, and the calculation method is: Where tf(t) is the TF value of the tth word segmentation after correction in the current remote communication process, W t is the weight coefficient of the t-th segmentation in the current remote communication process, f(t) is the frequency of occurrence of the t-th segmentation in the current remote communication process, and m is the number of segmentations in the current remote communication process.

9. The remote communication-based on-hook message adaptive push system according to claim 1, characterized in that: The filtering of each word by the TF-IDF value includes: Obtaining each segmentation that does not appear in the current telecommunication process but appears in the historical telecommunication process as a historical segmentation, and multiplying the TF-IDF value of each historical segmentation by a preset attenuation coefficient as the relative importance of each historical segmentation; The TF-IDF values of all the segmented words in the current remote communication process and the relative importance of all the historical segmented words are arranged in descending order, and a preset number of segmented words that are ranked at the top are used as the segmented words obtained by screening.

10. The remote communication-based on-hook message adaptive push system according to claim 9, characterized in that: The above is matched with the recommended information, and the idle message is pushed, including: The similarity between the filtered segmented words and all segmented words of each on-hook message in the on-hook message push library is calculated, and the on-hook message with the highest similarity is pushed.

Images