Text subject recognition method, device and equipment and storage medium

Abstract

The present application relates to artificial intelligence, and provides a text topic identification method, device, equipment and storage medium. The method identifies a target keyword based on a first keyword in a first text in a monitoring period and a second keyword in a second text in a previous period, calculates a first word frequency of the target keyword in the first text and a second word frequency of the target keyword in the second text, quantifies a word frequency proportion fluctuation based on the first word frequency and the second word frequency, generates a fluctuation credibility according to a relationship between a weighted word frequency proportion generated by the first word frequency and the second word frequency and a preset inflection point parameter, identifies a fluctuation index based on the word frequency proportion fluctuation and the fluctuation credibility, selects a candidate vocabulary according to the fluctuation index, and performs clustering processing on the candidate vocabulary to improve the accuracy of the topic information. In addition, the present application also relates to blockchain technology, and the topic information can be stored in the blockchain.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular to a text topic identification method, device, equipment and storage medium. BACKGROUND

[0002] In a telephone service scenario, whether the incoming call is related to a certain public opinion / special event can be identified through the incoming text data. At present, since the customer usually does not repeatedly emphasize or explain the public opinion / special event to the agent in the interaction process, the data topic of the incoming text data cannot be identified, resulting in serious lag in event response, and even missing.

[0003] Therefore, how to identify the topic information of the incoming text data to assist the identification of public opinion events has become a technical problem to be solved. SUMMARY

[0004] In view of the above, it is necessary to provide a text topic identification method, device, equipment and storage medium, which can solve the technical problem that the topic information of the incoming text data cannot be accurately identified.

[0005] In one aspect, the present application provides a text topic identification method, which comprises:

[0006] obtaining a first text of a monitoring period and a second text of a previous period;

[0007] identifying a target keyword based on a first keyword in the first text and a second keyword in the second text;

[0008] calculating a first word frequency of the target keyword in the first text and a second word frequency of the target keyword in the second text;

[0009] quantifying a word frequency proportion fluctuation of the target keyword based on the first word frequency and the second word frequency;

[0010] generating a fluctuation credibility of the target keyword according to the relationship between the weighted word frequency proportion generated by the first word frequency and the second word frequency and a preset inflection point parameter;

[0011] identifying a fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation credibility;

[0012] selecting a candidate vocabulary from the target keyword according to the fluctuation index and a preset fluctuation threshold;

[0013] performing clustering processing on the candidate vocabulary based on the first text to obtain the topic information of the first text.

[0014] According to a preferred embodiment of the present application, the identifying the target keyword based on the first keyword in the first text and the second keyword in the second text comprises:

[0015] normalizing the first text based on a preset identifier to obtain a first standard text, and normalizing the second text based on the preset identifier to obtain a second standard text;

[0016] performing word segmentation on the first standard text to obtain a first segmented word, and performing word segmentation on the second standard text to obtain a second segmented word;

[0017] cleaning the first segmented word based on a preset stop word list to obtain the first keyword, and cleaning the second segmented word based on the preset stop word list to obtain the second keyword;

[0018] determining the first keyword identical to the second keyword as the target keyword.

[0019] According to a preferred embodiment of the present application, the calculation formula of the word frequency proportion fluctuation is:

[0020]

[0021] wherein fct(P) represents the word frequency proportion fluctuation, freq t (P) represents the first word frequency, and freq c (P) represents the second word frequency.

[0022] According to a preferred embodiment of the present application, the generating the fluctuation credibility of the target keyword according to the relationship between the weighted word frequency proportion generated based on the first word frequency and the second word frequency and a preset inflection point parameter comprises:

[0023] performing weighted sum operation on the first word frequency and the second word frequency to obtain the weighted word frequency proportion;

[0024] comparing the weighted word frequency proportion with the preset inflection point parameter;

[0025] if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, determining the fluctuation credibility as a configuration value; or

[0026] if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, determining the ratio of the weighted word frequency proportion to the preset inflection point parameter as the fluctuation credibility.

[0027] According to a preferred embodiment of the present application, before generating the fluctuation credibility of the target keyword based on the relationship between the weighted word frequency ratio generated according to the first word frequency and the second word frequency and the preset inflection point parameter, the method further comprises:

[0028] obtaining a parameter range and obtaining training text;

[0029] obtaining a plurality of training inflection point parameters from the parameter range according to a preset ratio;

[0030] performing test processing on the training text based on the plurality of training inflection point parameters to obtain a test result;

[0031] determining the training inflection point parameter with the optimal test result as the preset inflection point parameter.

[0032] According to a preferred embodiment of the present application, the identifying the fluctuation index of the target keyword based on the word frequency ratio fluctuation and the fluctuation credibility comprises:

[0033] calculating the product of the word frequency ratio fluctuation and the fluctuation credibility to obtain the fluctuation index.

[0034] According to a preferred embodiment of the present application, the performing clustering processing on the candidate vocabulary based on the first text to obtain the theme information of the first text comprises:

[0035] determining any vocabulary in the candidate vocabulary as a theme center word and determining the candidate vocabulary other than the theme center word as a non-center word;

[0036] counting the number of center vocabulary of the theme center word in the first text and counting the number of non-center vocabulary of the non-center word in the first text;

[0037] determining the ratio of the number of center vocabulary to the number of non-center vocabulary as the co-occurrence rate of the theme center word and the non-center word;

[0038] merging the non-center word with the co-occurrence rate greater than a preset co-occurrence threshold to the theme of the theme center word to obtain a plurality of initial themes;

[0039] merging the themes containing the same vocabulary in the plurality of initial themes to obtain the theme information.

[0040] On the other hand, the present application also proposes a text theme recognition device, which comprises:

[0041] an obtaining unit, configured to obtain a first text in a monitoring period and a second text in a previous period;

[0042] An identifying unit is configured to identify a target keyword based on a first keyword in the first text and a second keyword in the second text.

[0043] A calculating unit is configured to calculate a first word frequency of the target keyword in the first text and a second word frequency of the target keyword in the second text.

[0044] A quantifying unit is configured to quantify a word frequency proportion fluctuation of the target keyword based on the first word frequency and the second word frequency.

[0045] A generating unit is configured to generate a fluctuation credibility of the target keyword according to a relationship between a weighted word frequency proportion generated from the first word frequency and the second word frequency and a preset inflection point parameter.

[0046] The identifying unit is further configured to identify a fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation credibility.

[0047] A selecting unit is configured to select a candidate vocabulary from the target keyword according to the fluctuation index and a preset fluctuation threshold.

[0048] A clustering unit is configured to perform clustering processing on the candidate vocabulary based on the first text to obtain theme information of the first text.

[0049] In another aspect, the present application further provides an electronic device, which comprises:

[0050] a memory configured to store computer readable instructions; and

[0051] a processor configured to execute the computer readable instructions stored in the memory to implement the text theme identification method.

[0052] In another aspect, the present application further provides a computer readable storage medium, which stores computer readable instructions, and the computer readable instructions are executed by a processor in an electronic device to implement the text theme identification method.

[0053] As can be seen from the above technical solutions, by taking a previous period as a comparison period, the present application can ensure that the first text and the second text are as similar as possible in structure, so as to reduce the interference of regular in-line texts when extracting abnormal change features of a monitoring period, and further, the word frequency proportion fluctuation and the fluctuation credibility can be combined to accurately identify the fluctuation index of the target keyword, and then the candidate vocabulary is screened out through the fluctuation index for clustering, which effectively reduces the interference of non-event low-frequency words on the candidate vocabulary, while retaining the possibility of low-frequency high-fluctuation event characteristic words being warned, and improves the identification accuracy of the theme information. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 is a flowchart of a preferred embodiment of the text subject identification method of the present application.

[0055] Figure 2 is a functional module diagram of a preferred embodiment of the text subject identification device of the present application.

[0056] Figure 3 is a structural schematic diagram of an electronic device implementing a preferred embodiment of the text subject identification method of the present application. DETAILED DESCRIPTION

[0057] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] As shown in Figure 1 is a flowchart of a preferred embodiment of the text subject identification method of the present application. The order of the steps in the flowchart can be changed according to different needs, and some steps can be omitted.

[0059] The text subject identification method can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) is the use of digital computers or computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0060] Artificial intelligence basic technologies generally include technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. Artificial intelligence software technologies mainly include computer vision technology, robot technology, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0061] The text subject identification method is applied in one or more electronic devices, which is a device capable of automatically performing numerical calculation and / or information processing according to pre-set or stored computer readable instructions. Its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0062] The electronic device can be any electronic product capable of human-computer interaction with a user, such as a personal computer, a tablet computer, a smart phone, a personal digital assistant (PDA), a game console, an interactive Internet Protocol Television (IPTV), a smart wearable device, or the like.

[0063] The electronic device can include a network device and / or a user device. The network device includes, but is not limited to, a single network electronic device, a group of electronic devices composed of multiple network electronic devices, or a cloud composed of a large number of hosts or network electronic devices based on cloud computing.

[0064] The network in which the electronic device is located includes, but is not limited to, the Internet, a wide area network, a metropolitan area network, a local area network, a virtual private network (VPN), or the like.

[0065] 101, obtaining first text in a monitoring period and second text in a pre-period.

[0066] In at least one embodiment of the present application, the monitoring period refers to a period during customer service in a telephone service scenario, for example, the monitoring period can be 9:00-10:00. The pre-period has a length equal to that of the monitoring period, and the pre-period is adjacent to the monitoring period, for example, if the length of the monitoring period is 2 hours, the pre-period is 2 hours before the monitoring period.

[0067] The first text refers to incoming text data in the monitoring period, and the second text refers to incoming text data in the pre-period, wherein the incoming text data refers to a call record text between a user and a customer service in a telephone service scenario.

[0068] In at least one embodiment of the present application, the electronic device obtaining the first text in the monitoring period and the second text in the pre-period includes:

[0069] The first call record in the monitoring period is collected by a buried point, and the second call record in the pre-period is collected by the buried point;

[0070] The first call record is text-converted to obtain the first text, and the second call record is text-converted to obtain the second text.

[0071] 102, identifying a target keyword based on a first keyword in the first text and a second keyword in the second text.

[0072] In at least one embodiment of the present application, the first keyword refers to a word obtained after preprocessing the first text, the second keyword refers to a word obtained after preprocessing the second text, and the target keyword refers to a word identical to the first keyword and the second keyword.

[0073] In at least one embodiment of the present application, the electronic device identifies the target keyword based on a first keyword in the first text and a second keyword in the second text, which includes:

[0074] regularizing the first text based on a preset identifier to obtain a first standard text, and regularizing the second text based on the preset identifier to obtain a second standard text;

[0075] performing word segmentation on the first standard text to obtain a first segmented word, and performing word segmentation on the second standard text to obtain a second segmented word;

[0076] cleaning the first segmented word based on a preset stop word table to obtain the first keyword, and cleaning the second segmented word based on the preset stop word table to obtain the second keyword;

[0077] determining the first keyword identical to the second keyword as the target keyword.

[0078] The preset identifier can be any symbol mark.

[0079] The first standard text refers to text information obtained by normalizing numbers and locations in the first text using the preset identifier.

[0080] The preset stop word table stores a plurality of words without semantic information, for example, the preset identifier is stored in the preset stop word table.

[0081] Through the above implementation, the interference of words without semantic information can be avoided, and the identification accuracy of the target keyword is improved.

[0082] 103, calculating a first word frequency of the target keyword in the first text and a second word frequency of the target keyword in the second text.

[0083] In at least one embodiment of the present application, the first word frequency refers to the frequency of the target keyword appearing in the first text, and the second word frequency refers to the frequency of the target keyword appearing in the second text.

[0084] In at least one embodiment of the present application, the electronic device calculates the first word frequency of the target keyword in the first text, which includes:

[0085] counting a number of words of the target keyword in the first text and counting a number of texts of the first text containing the target keyword;

[0086] calculating a ratio of the number of words to the number of texts to obtain the first word frequency.

[0087] In at least one embodiment of the present application, the second word frequency is calculated in a similar manner to the first word frequency, which will not be described herein.

[0088] 104, quantifying a word frequency proportion fluctuation of the target keyword based on the first word frequency and the second word frequency.

[0089] In at least one embodiment of the present application, the word frequency proportion fluctuation refers to a fluctuation of the target keyword in the first text and the second text.

[0090] In at least one embodiment of the present application, the calculation formula of the word frequency proportion fluctuation is:

[0091]

[0092] wherein fct(P) represents the word frequency proportion fluctuation, freq t (P) represents the first word frequency, freq c (P) represents the second word frequency.

[0093] Through the above implementation, the word frequency proportion fluctuation can be accurately quantified.

[0094] 105, generating a fluctuation credibility of the target keyword according to a relationship between a weighted word frequency proportion generated based on the first word frequency and the second word frequency and a preset inflection point parameter.

[0095] In at least one embodiment of the present application, the weighted word frequency proportion is obtained by weighting and operating the first word frequency and the second word frequency according to a text weight of the first text and a text weight of the second text. It should be noted that the text weight of each target keyword in the first text is the same.

[0096] The preset inflection point parameter can be trained according to training texts.

[0097] In at least one embodiment of the present application, before generating the fluctuation credibility of the target keyword according to the relationship between the weighted word frequency proportion generated based on the first word frequency and the second word frequency and the preset inflection point parameter, the method further comprises:

[0098] obtaining a parameter range and obtaining training texts;

[0099] acquire a plurality of training inflection point parameters from the parameter range according to a preset ratio;

[0100] perform test processing on the training text based on the plurality of training inflection point parameters to obtain a test result;

[0101] determine the training inflection point parameter with the optimal test result as the preset inflection point parameter.

[0102] The parameter range can be a pre-set data interval.

[0103] The preset ratio can be determined according to the recognition accuracy of the preset inflection point parameter.

[0104] The test result can represent the training ability of each training inflection point parameter for generating the training text, for example, the test result can be a loss value.

[0105] The preset inflection point parameter refers to the training inflection point parameter with the minimum loss value.

[0106] By performing a grid search on the preset inflection point parameter, the accuracy of the preset inflection point parameter can be improved, thereby improving the accuracy of the fluctuation confidence.

[0107] In at least one embodiment of the present application, the electronic device generates the fluctuation confidence of the target keyword according to the relationship between the weighted word frequency proportion generated by the first word frequency and the second word frequency and the preset inflection point parameter, which includes:

[0108] performing weighted sum operation on the first word frequency and the second word frequency to obtain the weighted word frequency proportion;

[0109] comparing the weighted word frequency proportion with the preset inflection point parameter;

[0110] if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, the fluctuation confidence is determined as a configuration value; or

[0111] if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, the ratio of the weighted word frequency proportion to the preset inflection point parameter is determined as the fluctuation confidence.

[0112] Through the above embodiments, the influence of the first text and the second text on the theme recognition quality can be retained at the same time.

[0113] 106, identify the fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation confidence.

[0114] In at least one embodiment of the present application, the electronic device identifies a fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation credibility, which includes:

[0115] The product of the word frequency proportion fluctuation and the fluctuation credibility is calculated to obtain the fluctuation index.

[0116] By combining the word frequency proportion fluctuation and the fluctuation credibility to calculate the fluctuation index, the influence of ultra-low frequency words on theme identification due to recognition errors in text conversion can be avoided.

[0117] 107. Select a candidate vocabulary from the target keyword according to the fluctuation index and a preset fluctuation threshold.

[0118] In at least one embodiment of the present application, the preset fluctuation threshold can be set according to actual needs, for example, the preset fluctuation threshold can be set to 200.

[0119] In at least one embodiment of the present application, the candidate vocabulary refers to the target keyword whose fluctuation index is greater than the preset fluctuation threshold.

[0120] 108. Cluster processing the candidate vocabulary based on the first text to obtain theme information of the first text.

[0121] It should be emphasized that, in order to further ensure the privacy and security of the above-mentioned theme information, the above-mentioned theme information can also be stored in a node of a block chain.

[0122] In at least one embodiment of the present application, the theme information is used to represent the main idea of the first text. The theme information can be used to process tasks such as mining customer complaint feature expression, customer cancellation scenario high retention rate dialogue, long pass time call features, etc.

[0123] In at least one embodiment of the present application, the electronic device clusters the candidate vocabulary based on the first text to obtain theme information of the first text, which includes:

[0124] Any vocabulary in the candidate vocabulary is determined as a theme center word, and the candidate vocabulary other than the theme center word is determined as a non-center word;

[0125] The number of center words of the theme center word in the first text is counted, and the number of non-center words of the non-center word in the first text is counted;

[0126] The ratio of the number of center words to the number of non-center words is determined as the co-occurrence rate of the theme center word and the non-center word;

[0127] merge the non-center words with the co-occurrence rate greater than the preset co-occurrence threshold to the theme of the theme center word to obtain a plurality of initial themes;

[0128] merge the themes containing the same words in the plurality of initial themes to obtain the theme information.

[0129] The theme of the theme center word refers to a set containing the theme center word.

[0130] By merging the themes with the same words, the theme information can be made more reasonable, and by constructing the theme based on the keywords filtered from the first text, the accuracy of the theme information can be improved.

[0131] From the above technical solutions, it can be seen that by taking the preceding period as the comparison period, the first text and the second text can be made as similar as possible in structure, so as to reduce the interference of the regular incoming text when extracting the abnormal change features of the monitoring period. Further, the fluctuation index of the target keyword can be accurately identified by combining the word frequency proportion fluctuation and the fluctuation credibility, and then the to-be-selected words are filtered and clustered by the fluctuation index, which effectively reduces the interference of non-event low-frequency words on the to-be-selected words, and also retains the possibility of low-frequency high-fluctuation event feature words being warned, thereby improving the identification accuracy of the theme information.

[0132] As shown in Figure 2 Fig. 1 is a functional module diagram of a preferred embodiment of a text theme identification device according to the present application. The text theme identification device 11 comprises an acquisition unit 110, an identification unit 111, a calculation unit 112, a quantization unit 113, a generation unit 114, a selection unit 115, a clustering unit 116, a test unit 117, and a determination unit 118. The modules / units referred to in the present application are a series of computer-readable instruction segments capable of being acquired by a processor 13 and capable of completing a fixed function, which are stored in a memory 12. In the present embodiment, the functions of the modules / units will be described in detail in subsequent embodiments.

[0133] The acquisition unit 110 acquires a first text of a monitoring period and a second text of a preceding period.

[0134] In at least one embodiment of the present application, the monitoring period refers to a period during customer service in a telephone service scenario, for example, the monitoring period can be 9:00-10:00. The length of the preceding period is equal to the length of the monitoring period, and the preceding period is adjacent to the monitoring period, for example, if the length of the monitoring period is 2 hours, the preceding period is 2 hours before the monitoring period.

[0135] The first text refers to in-line text data in the monitoring period, and the second text refers to in-line text data in the prelude period, wherein the in-line text data refers to a call record text between a user and a customer service in a telephone service scenario.

[0136] In at least one embodiment of the present application, the acquisition unit 110 acquires the first text in the monitoring period and the second text in the prelude period, including:

[0137] The in-line text data in the monitoring period is collected as the first call record, and the in-line text data in the prelude period is collected as the second call record.

[0138] The first call record is text-converted to obtain the first text, and the second call record is text-converted to obtain the second text.

[0139] The identification unit 111 identifies a target keyword based on a first keyword in the first text and a second keyword in the second text.

[0140] In at least one embodiment of the present application, the first keyword refers to a vocabulary obtained after preprocessing of the first text, the second keyword refers to a vocabulary obtained after preprocessing of the second text, and the target keyword refers to a vocabulary identical to the first keyword and the second keyword.

[0141] In at least one embodiment of the present application, the identification unit 111 identifies a target keyword based on a first keyword in the first text and a second keyword in the second text, including:

[0142] The first text is regularized based on a preset identifier to obtain a first standard text, and the second text is regularized based on the preset identifier to obtain a second standard text;

[0143] The first standard text is processed by word segmentation to obtain a first segmented word, and the second standard text is processed by word segmentation to obtain a second segmented word;

[0144] The first segmented word is cleaned based on a preset stop word list to obtain the first keyword, and the second segmented word is cleaned based on the preset stop word list to obtain the second keyword;

[0145] The first keyword identical to the second keyword is determined as the target keyword.

[0146] The preset identifier can be any symbol mark.

[0147] The first standard text refers to text information obtained by normalizing numbers and locations in the first text by using the preset identifier.

[0148] The preset stop word list stores a plurality of words without semantic information, for example, the preset identifier.

[0149] Through the above implementation, the interference of words without semantic information can be avoided, and the recognition accuracy of the target keyword is improved.

[0150] The computing unit 112 calculates a first word frequency of the target keyword in the first text and a second word frequency of the target keyword in the second text.

[0151] In at least one embodiment of the present application, the first word frequency refers to the frequency of the target keyword appearing in the first text, and the second word frequency refers to the frequency of the target keyword appearing in the second text.

[0152] In at least one embodiment of the present application, the computing unit 112 calculates the first word frequency of the target keyword in the first text, comprising:

[0153] Counting the number of words of the target keyword in the first text, and counting the number of texts containing the target keyword in the first text;

[0154] Calculating the ratio of the number of words to the number of texts to obtain the first word frequency.

[0155] In at least one embodiment of the present application, the second word frequency is calculated in a similar manner to the first word frequency, which will not be described herein.

[0156] The quantifying unit 113 quantifies the word frequency proportion fluctuation of the target keyword based on the first word frequency and the second word frequency.

[0157] In at least one embodiment of the present application, the word frequency proportion fluctuation refers to the fluctuation of the target keyword in the first text and the second text.

[0158] In at least one embodiment of the present application, the calculation formula of the word frequency proportion fluctuation is:

[0159]

[0160] Wherein, fct(P) represents the word frequency proportion fluctuation, freq t (P) represents the first word frequency, freq c (P) represents the second word frequency.

[0161] By the above-mentioned embodiments, the word frequency proportion fluctuation of the target keyword can be accurately quantified.

[0162] The generation unit 114 generates the fluctuation credibility of the target keyword according to the relationship between the weighted word frequency proportion generated according to the first word frequency and the second word frequency and the preset inflection point parameter.

[0163] In at least one embodiment of the present application, the weighted word frequency proportion is obtained by weighting and adding the first word frequency and the second word frequency according to the text weight of the first text and the text weight of the second text. It should be noted that the text weight of each target keyword in the first text is the same.

[0164] The preset inflection point parameter can be trained according to training text.

[0165] In at least one embodiment of the present application, before the fluctuation credibility of the target keyword is generated according to the relationship between the weighted word frequency proportion generated according to the first word frequency and the second word frequency and the preset inflection point parameter, the acquisition unit 110 acquires a parameter range and acquires training text.

[0166] The acquisition unit 110 acquires a plurality of training inflection point parameters from the parameter range according to a preset ratio;

[0167] The test unit 117 performs test processing on the training text based on the plurality of training inflection point parameters to obtain a test result.

[0168] The determination unit 118 determines the training inflection point parameter with the optimal test result as the preset inflection point parameter.

[0169] The parameter range can be a pre-set data interval.

[0170] The preset ratio can be determined according to the recognition accuracy of the preset inflection point parameter.

[0171] The test result can represent the training ability of each training inflection point parameter for generating the training text, for example, the test result can be a loss value.

[0172] The preset inflection point parameter refers to the training inflection point parameter with the minimum loss value.

[0173] By grid searching the preset inflection point parameter, the accuracy of the preset inflection point parameter can be improved, thereby improving the accuracy of the fluctuation credibility.

[0174] In at least one embodiment of the present application, the generating unit 114 generates the fluctuation confidence of the target keyword according to the relationship between the weighted word frequency proportion generated according to the first word frequency and the second word frequency and the preset inflection point parameter, which comprises:

[0175] performing weighted sum operation on the first word frequency and the second word frequency to obtain the weighted word frequency proportion;

[0176] comparing the weighted word frequency proportion with the preset inflection point parameter;

[0177] if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, determining the fluctuation confidence as a configuration value; or

[0178] if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, determining the ratio of the weighted word frequency proportion to the preset inflection point parameter as the fluctuation confidence.

[0179] Through the above-mentioned embodiments, the influence of the first text and the second text on the theme recognition quality can be reserved at the same time.

[0180] The recognition unit 111 identifies the fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation confidence.

[0181] In at least one embodiment of the present application, the recognition unit 111 identifies the fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation confidence, which comprises:

[0182] calculating the product of the word frequency proportion fluctuation and the fluctuation confidence to obtain the fluctuation index.

[0183] By calculating the fluctuation index by combining the word frequency proportion fluctuation and the fluctuation confidence, the influence of the ultra-low frequency words on the theme recognition due to the recognition error of the text conversion can be avoided.

[0184] The selection unit 115 selects the candidate word from the target keyword according to the fluctuation index and a preset fluctuation threshold.

[0185] In at least one embodiment of the present application, the preset fluctuation threshold can be set according to actual needs, for example, the preset fluctuation threshold can be set to 200.

[0186] In at least one embodiment of the present application, the candidate word refers to the target keyword whose fluctuation index is greater than the preset fluctuation threshold.

[0187] The clustering unit 116 performs clustering processing on the candidate word based on the first text to obtain the theme information of the first text.

[0188] It is emphasized that, in order to further ensure the privacy and security of the above-mentioned subject information, the above-mentioned subject information can also be stored in a node of a block chain.

[0189] In at least one embodiment of the present application, the subject information is used to characterize the main idea of the first text. The subject information can be used to process tasks such as mining customer complaint feature expressions, high retention rate scripts in customer cancellation scenarios, and long call time call features.

[0190] In at least one embodiment of the present application, the clustering unit 116 performs clustering processing on the candidate vocabulary based on the first text to obtain the subject information of the first text, including:

[0191] Any of the candidate vocabulary is determined as a subject center word, and the candidate vocabulary other than the subject center word is determined as a non-center word;

[0192] The number of center words of the subject center word in the first text is counted, and the number of non-center words of the non-center word in the first text is counted;

[0193] The ratio of the number of center words to the number of non-center words is determined as the co-occurrence rate of the subject center word and the non-center word;

[0194] The non-center word with a co-occurrence rate greater than a preset co-occurrence threshold is merged into the subject of the subject center word to obtain a plurality of initial subjects;

[0195] The subjects containing the same words in the plurality of initial subjects are merged to obtain the subject information.

[0196] Wherein, the subject of the subject center word refers to a set containing the subject center word.

[0197] By merging the subjects with the same words, the subject information can be avoided to be general and not precise, thereby improving the rationality of the subject information. At the same time, constructing the subject based on the keywords filtered from the first text can improve the accuracy of the subject information.

[0198] From the above technical scheme can be seen, the present application by obtaining the prelude period as a comparison period, can ensure that the first text and the second text in structure as possible similar, in order to reduce the interference generated by the conventional incoming line text when extracting monitoring period abnormal change characteristics, further, in combination with the word frequency proportion fluctuation and the fluctuation credibility can accurately identify the target keyword fluctuation index, and then through the fluctuation index screening out the candidate vocabulary clustering, effectively reduces the non event low frequency word caused by the interference of the candidate vocabulary, at the same time also retains the possibility of low frequency high fluctuation event characteristic word is early warning, improve the identification accuracy of the theme information.

[0199] As Figure 3 shown, it is the structure schematic diagram of electronic equipment of preferred embodiment of the present application for realizing text theme identification method.

[0200] In an embodiment of the present application, the electronic equipment 1 includes, but is not limited to, a memory 12, a processor 13, and computer readable instructions stored in the memory 12 and executable on the processor 13, such as a text theme identification program.

[0201] Those skilled in the art can understand that the schematic diagram is only an example of the electronic equipment 1, and does not constitute a limitation on the electronic equipment 1, and can include more or less components than the diagram, or combine certain components, or different components, for example, the electronic equipment 1 can also include input / output devices, network access devices, buses, etc.

[0202] The processor 13 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The processor 13 is the operation core and control center of the electronic equipment 1, and connects various parts of the entire electronic equipment 1 through various interfaces and lines, and executes the operating system and installed various application programs, program codes, etc. of the electronic equipment 1.

[0203] For example, the computer readable instructions can be divided into one or more modules / units, which are stored in the memory 12 and executed by the processor 13 to complete the present application. The one or more modules / units can be a series of computer readable instruction segments capable of completing a specific function, which are used to describe the execution process of the computer readable instructions in the electronic device 1. For example, the computer readable instructions can be divided into an acquisition unit 110, an identification unit 111, a calculation unit 112, a quantization unit 113, a generation unit 114, a selection unit 115, a clustering unit 116, a test unit 117, and a determination unit 118.

[0204] The memory 12 can be used to store the computer readable instructions and / or modules, and the processor 13 realizes various functions of the electronic device 1 by running or executing the computer readable instructions and / or modules stored in the memory 12, and calling the data stored in the memory 12. The memory 12 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, an image playing function, etc.), and the like; the data storage area can store data created according to the use of the electronic device, etc. The memory 12 can include non-volatile and volatile memories, for example: a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash storage device, or other storage devices.

[0205] The memory 12 can be an external memory and / or an internal memory of the electronic device 1. Further, the memory 12 can be a memory having a physical form, such as a memory stick, a TF card (Trans-flash Card), and the like.

[0206] The modules / units integrated in the electronic device 1, if implemented in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by computer readable instructions instructing related hardware, and the computer readable instructions can be stored in a computer readable storage medium. When the computer readable instructions are executed by a processor, the steps of the above-mentioned various method embodiments can be implemented.

[0207] The computer readable instructions include computer readable instruction codes in source code form, object code form, executable files, or some intermediate form, etc. The computer readable medium can include any entity or device capable of carrying the computer readable instruction codes, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM).

[0208] The blockchain referred to in the present application is a new application mode of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm. Blockchain, in essence, is a decentralized database, which is a series of data blocks associated using cryptographic methods, each of which contains information about a batch of network transactions, used to verify the validity of the information (anti-fake) and generate the next block. The blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer.

[0209] In combination Figure 1 , the memory 12 in the electronic device 1 stores computer readable instructions to implement a text topic identification method, and the processor 13 can execute the computer readable instructions to implement:

[0210] Obtain a first text in a monitoring period and a second text in a previous period;

[0211] Identify a target keyword based on a first keyword in the first text and a second keyword in the second text;

[0212] Calculate a first word frequency of the target keyword in the first text and a second word frequency of the target keyword in the second text;

[0213] Quantify the word frequency proportion fluctuation of the target keyword based on the first word frequency and the second word frequency;

[0214] Generate the fluctuation credibility of the target keyword according to the relationship between the weighted word frequency proportion generated by the first word frequency and the second word frequency and a preset inflection point parameter;

[0215] Identify the fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation credibility;

[0216] Select a candidate vocabulary from the target keyword according to the fluctuation index and a preset fluctuation threshold;

[0217] Cluster the candidate vocabulary based on the first text to obtain the topic information of the first text.

[0218] Specifically, the processor 13 can refer to the specific implementation method of the above computer readable instructions Figure 1 The description of the related steps in the corresponding embodiment will not be repeated here.

[0219] In several embodiments provided by the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. In actual implementation, there can be another division manner.

[0220] The computer readable storage medium stores computer readable instructions, wherein the computer readable instructions are executed by the processor 13 to implement the following steps:

[0221] Obtain the first text in the monitoring period and the second text in the previous period;

[0222] Identify the target keyword based on the first keyword in the first text and the second keyword in the second text;

[0223] Calculate the first word frequency of the target keyword in the first text, and calculate the second word frequency of the target keyword in the second text;

[0224] Quantify the word frequency proportion fluctuation of the target keyword based on the first word frequency and the second word frequency;

[0225] Generate the fluctuation credibility of the target keyword according to the relationship between the weighted word frequency proportion generated by the first word frequency and the second word frequency and the preset inflection point parameter;

[0226] Identify the fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation credibility;

[0227] Select the candidate vocabulary from the target keyword according to the fluctuation index and the preset fluctuation threshold;

[0228] Cluster the candidate vocabulary based on the first text to obtain the theme information of the first text.

[0229] The modules described as separate components can or can not be physically separated, and the components displayed as modules can or can not be physical units, that is, they can be located in one place, or they can be distributed on multiple network units. According to actual needs, some or all of the modules can be selected to achieve the purpose of the present embodiment.

[0230] In addition, each function module in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware, or in the form of hardware plus software function module.

[0231] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, the scope of the present application is defined by the appended claims rather than the above description, and therefore all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application. Any additional reference signs in the claims should not be considered as limiting the claims involved.

[0232] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The plurality of units or devices mentioned can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, not any particular order.

[0233] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A method of text topic identification, characterized by, The text theme recognition method comprises: acquiring first text of a monitoring period and second text of a previous period; identifying a target keyword based on a first keyword in the first text and a second keyword in the second text; calculating a first word frequency of the target keyword in the first text and a second word frequency of the target keyword in the second text; Quantify the word frequency proportion fluctuation of the target keyword based on the first word frequency and the second word frequency, and the calculation formula of the word frequency proportion fluctuation is: ; wherein, represents the word frequency proportion fluctuation, represents the first word frequency, represents the second word frequency; generating a fluctuation credibility of the target keyword according to a relationship between a weighted word frequency proportion generated from the first word frequency and the second word frequency and a preset inflection point parameter, comprising: performing weighted sum operation on the first word frequency and the second word frequency to obtain the weighted word frequency proportion; comparing the weighted word frequency proportion with the preset inflection point parameter; if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, determining the fluctuation credibility as a configuration value; or if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, determining a ratio of the weighted word frequency proportion to the preset inflection point parameter as the fluctuation credibility; identifying a fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation credibility, comprising: calculating a product of the word frequency proportion fluctuation and the fluctuation credibility to obtain the fluctuation index; selecting a candidate vocabulary from the target keyword according to the fluctuation index and a preset fluctuation threshold; performing clustering processing on the candidate vocabulary based on the first text to obtain theme information of the first text.

2. The text topic identification method of claim 1, wherein, The method further comprises, before generating the fluctuation credibility of the target keyword according to the relationship between the weighted word frequency proportion generated from the first word frequency and the second word frequency and the preset inflection point parameter: acquiring a parameter range and acquiring training text; acquiring a plurality of training inflection point parameters from the parameter range according to a preset ratio; performing test processing on the training text based on the plurality of training inflection point parameters to obtain a test result; determining the training inflection point parameter with the optimal test result as the preset inflection point parameter.

3. The text topic identification method of claim 1, wherein, The method further comprises, before performing clustering processing on the candidate vocabulary based on the first text to obtain theme information of the first text: determining any vocabulary in the candidate vocabulary as a theme center word and determining the candidate vocabulary other than the theme center word as a non-center word. ​ ​ ​ 4. The text topic identification method of claim 1, wherein, ​ ​ count a number of central words of the topic central word in the first text and count a number of non-central words of the non-central word in the first text; determine a co-occurrence rate of the topic central word and the non-central word as a ratio of the number of central words to the number of non-central words; merge the non-central word with the co-occurrence rate greater than a preset co-occurrence threshold to a topic of the topic central word to obtain a plurality of initial topics; merge topics containing the same word in the plurality of initial topics to obtain the topic information.

5. A text topic identification apparatus characterized by comprising: The text topic identification device comprises: an acquisition unit configured to acquire a first text of a monitoring period and a second text of a previous period; an identification unit configured to identify a target keyword based on a first keyword in the first text and a second keyword in the second text; a calculation unit configured to calculate a first word frequency of the target keyword in the first text and calculate a second word frequency of the target keyword in the second text; A quantization unit is configured to quantize a word frequency proportion fluctuation of the target keyword based on the first word frequency and the second word frequency, and a calculation formula of the word frequency proportion fluctuation is: ; wherein, represents the word frequency proportion fluctuation, represents the first word frequency, represents the second word frequency; a generation unit configured to generate a fluctuation credibility of the target keyword according to a relationship between a weighted word frequency proportion generated from the first word frequency and the second word frequency and a preset inflection point parameter, including: performing weighted sum operation on the first word frequency and the second word frequency to obtain the weighted word frequency proportion; comparing the weighted word frequency proportion with the preset inflection point parameter; if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, determining the fluctuation credibility as a configuration value; or if the weighted word frequency proportion is greater than or equal to the preset inflection point parameter, determining a ratio of the weighted word frequency proportion to the preset inflection point parameter as the fluctuation credibility; the identification unit is further configured to identify a fluctuation index of the target keyword based on the word frequency proportion fluctuation and the fluctuation credibility, including: calculating a product of the word frequency proportion fluctuation and the fluctuation credibility to obtain the fluctuation index; a selection unit configured to select a to-be-selected word from the target keyword according to the fluctuation index and a preset fluctuation threshold; a clustering unit configured to perform clustering processing on the to-be-selected word based on the first text to obtain topic information of the first text.

6. An electronic device, comprising: The electronic device comprises: a memory storing computer readable instructions; and a processor executing the computer readable instructions stored in the memory to implement the text topic identification method of any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: The computer readable storage medium stores computer readable instructions, and the computer readable instructions are executed by the processor in the electronic device to implement the text topic identification method of any one of claims 1 to 4.

Images