A network public opinion information monitoring and processing method and system

Abstract

The present invention provides a network public opinion information monitoring and processing method and system, which belongs to the technical field of network public opinion information monitoring, and specifically comprises: obtaining network public opinion information to be monitored, extracting publishers of the network public opinion information, and constructing a monitor database based on historical publishing information of account subjects of a network public opinion information monitoring platform; obtaining key publishers based on matching publishers with the monitor database; performing word segmentation processing on the publishing information of the key publishers and matching it with a keyword word library to obtain the number of basic negative word segments, the number of negative English word segments, the number of negative pinyin word segments, the number of negative pinyin initials abbreviation word segments, and the number of English initials abbreviation word segments; based on the proportion of key publishers in publishers, the above-mentioned number of negative word segments is sent to a public opinion prediction model, and a public opinion judgment result is output, thereby ensuring the efficiency requirement and further improving the accuracy of recognition.

Description

Technical Field

[0001] The present invention belongs to the technical field of network public opinion information monitoring, and in particular relates to a network public opinion information monitoring and processing method and system. Background Art

[0002] Today, as network technology matures, the gradual development and popularization of Internet technology, especially mobile Internet technology, has integrated the Internet into all aspects of social life. With the lowering of the threshold for online social networking, netizens who previously could only search and obtain the information they needed from the Internet can now participate in the exchange and creation of online information by creating their own social media or logging into numerous online social platforms. A research report by the China Internet Information Center pointed out that as of June this year, the number of Chinese netizens has exceeded 1 billion.

[0003] At the same time, due to the existence of intentional spelling errors or a large number of abbreviated expressions on the Internet, inaccurate results may occur during the identification of online public opinion information. In order to solve the above problems, China's authorized invention patent publication number CN103544242B "A sentiment entity search system for Weibo" uses a constructed word relationship graph to expand queries and realize the identification of typos or abbreviated expressions. However, similar spelling errors or a large number of abbreviated expressions are often updated very quickly on the Internet, and there are more negative emotional components in public opinion information with similar spelling errors. Therefore, if targeted identification and expansion cannot be performed according to the common types of spelling errors, and supplementary verification is performed based on the publisher's historical information release, a large recognition error rate will be caused, and the results of online public opinion monitoring will be inaccurate, so that the public opinion situation cannot be paid attention to in the first time, and once a risk occurs, it will bring serious consequences.

[0004] In response to the above technical problems, the present invention provides a method and system for monitoring and processing network public opinion information. Summary of the invention

[0005] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:

[0006] According to one aspect of the present invention, a method for monitoring and processing network public opinion information is provided.

[0007] A method for monitoring and processing network public opinion information, characterized in that it specifically includes:

[0008] S11 obtains the network public opinion information to be monitored, extracts the publisher of the network public opinion information, and constructs a monitor database based on the historical publishing information of the account subject of the monitoring platform of the network public opinion information;

[0009] S12: matching the publisher subject with the monitor database, and obtaining a key publisher subject based on the matching result;

[0010] S13 extracts the published information of the key publisher, and performs word segmentation processing on the published information to obtain basic word segmentation;

[0011] S14 matches the basic segmentation words with the keyword word library to obtain the number of basic negative segmentation words; performs English translation based on the segmentation words other than the basic negative words to form English segmentation words, and compares the English segmentation words with the English keyword database to determine the number of negative English segmentation words, based on the segmentation words excluding negative English segmentation words and basic negative words, and matches the pinyin, pinyin abbreviation, and English abbreviation of the segmentation words excluding negative English segmentation words and basic negative words with the pinyin keyword database, the pinyin abbreviation database, and the English abbreviation database, respectively, to obtain the number of negative pinyin segmentation words, the number of negative pinyin abbreviation segmentation words, and the number of English abbreviation segmentation words;

[0012] S15 is based on the proportion of the key publishers among the publishers, the number of basic negative word segmentations, the number of negative English word segmentations, the number of negative pinyin word segmentations, the number of negative pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations, and inputs them into the public opinion prediction model based on the EPO-LSTM algorithm to output the public opinion judgment result.

[0013] First, the network public opinion information to be monitored is obtained, and the publisher subject is extracted, and a monitor database is constructed based on the historical release information, and the publisher subject is matched with the monitor database to find the key publisher subject. Since the key publisher is more likely to use typos or abbreviations, the key publishers are first screened, and based on the screening results, the release information of the key publishers is segmented, and the obtained segmented words are matched with the keyword thesaurus to obtain the number of basic negative segmented words, and the segmented words excluding the basic negative words are translated into English to form English segmented words, and the English segmented words are compared with the English keyword database to determine the number of negative English segmented words, based on the segmented words excluding negative English segmented words and basic negative words, and the pinyin, pinyin initials abbreviation, and English initials abbreviation of the segmented words excluding negative English segmented words and basic negative words are matched with the pinyin keyword database, the pinyin initials abbreviation database, and the English initials abbreviation database, respectively. , and obtain the number of negative pinyin segmentations, the number of negative pinyin abbreviation segmentations, and the number of English abbreviation segmentations, thereby solving the original inaccurate recognition problem caused by similar spelling errors or a large number of abbreviated expressions on the Internet, the update speed is very fast, and there are more negative emotional components in the public opinion information with similar spelling errors. By performing targeted recognition and expansion according to the common types of spelling errors, and supplementing verification based on the publisher's historical release information, the recognition result and recognition efficiency are greatly improved, further laying the foundation for realizing accurate judgment of public opinion information in the first time, and based on the proportion of the key publisher subject in the publisher, the basic negative segmentation number, the negative English segmentation number, the negative pinyin segmentation number, the negative pinyin abbreviation segmentation number, and the English abbreviation segmentation number are sent to the public opinion prediction model based on the EPO-LSTM algorithm to output the public opinion judgment result.

[0014] By establishing a monitor database based on historical information, a foundation is laid for further realizing the identification of key publishers, and further based on the matching results of the monitor database and the publisher, the key publishers are obtained, and the screening of people who often use typos and abbreviations is realized, thereby further improving the overall efficiency and accuracy. By matching the word segmentation results of the key publishers with the keyword database, the English keyword database, the pinyin keyword database, the pinyin abbreviation database, and the English abbreviation database, the basic negative word segmentation quantity is obtained, the negative English word segmentation quantity, the negative pinyin word segmentation quantity, the negative pinyin abbreviation word segmentation quantity, and the English abbreviation word segmentation quantity, thereby realizing the word segmentation quantity of negative words. The quantitative classification type matching further improves the accuracy of the recognition results and ensures the overall efficiency. The public opinion prediction model based on the EPO-LSTM algorithm is input based on the proportion of the key publishers, the number of basic negative word segmentations, the number of negative English word segmentations, the number of negative pinyin word segmentations, the number of negative pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations to output the public opinion judgment results, thereby solving the problem of the original statistical results based on all the people, which leads to low accuracy and lack of pertinence. It can achieve accurate judgment of public opinion, and take corresponding measures according to the results of public opinion judgment. The output public opinion judgment results are divided into five stages: germination period, rising period, outbreak period, stable period, and decline period.

[0015] A further technical solution is that the word segmentation method adopts the Jieba word segmentation model.

[0016] A further technical solution is that the specific steps of constructing the monitor database are:

[0017] S11: performing word segmentation processing on the historical release information of the account subject to obtain historical word segmentation, and matching the historical word segmentation with the keyword database to obtain negative historical word segmentation;

[0018] S12 obtains historical proportion data based on the proportion of the release information of the negative historical segmentation words in the historical distribution information;

[0019] S13: When the historical proportion data is greater than a first proportion threshold, the account subject is determined to be a monitor, and the monitor is placed in the monitor database.

[0020] By using the historical proportion data of the release information of negative historical segmentation words in the historical distribution information, it is determined whether the account subject is a monitor, thereby further achieving accurate positioning and analysis of the monitor. When the historical proportion data is greater than the first proportion threshold, it means that there is a lot of negative release information at this time, so that the monitor can be accurately located. At the same time, it also solves the one-sidedness of the original judgment based solely on quantity or words, so that the accuracy is enhanced to a certain extent.

[0021] A further technical solution is that the first percentage threshold is determined based on the number of account entities of the monitoring platform and the number of historical published information.

[0022] When the number of account entities is small and less than the first account threshold, the first proportion threshold can be set larger. When the number of account entities is greater than or equal to the first account threshold, it means that the number is large at this time, and the first proportion threshold can be lowered. When the number of historical published information is small and less than the first information threshold, the first proportion threshold can be set larger. When the number of historical published information is greater than or equal to the first information threshold, it means that the number is large at this time, and the first proportion threshold can be lowered. The setting of the first proportion threshold can be further realized according to the actual situation, so that the setting result becomes more reliable and applicable.

[0023] A further technical solution is that when the number of historical published information is less than a first quantity threshold and the historical proportion data is greater than a second proportion threshold, the account subject is determined to be a monitor; when the number of historical published information is greater than the second quantity threshold and the historical proportion data is greater than a third proportion threshold, the account subject is determined to be a monitor, the third proportion threshold is less than the first proportion threshold, the first proportion threshold is less than the second proportion threshold, and the first quantity threshold is less than the second quantity threshold.

[0024] The judgment of the monitor is made based on the number of historically published information of the account subject, thus solving the problem of misjudgment of account subjects with a small or large number of published information due to the use of a unified standard regardless of the number of historical published information, and further improving the judgment.

[0025] A further technical solution is to take the published information containing the basic negative words, the published information containing the number of negative English word segmentations, the published information containing the number of negative pinyin word segmentations, the published information containing the number of negative pinyin abbreviation word segmentations, and the published information containing the number of English abbreviation word segmentations as key negative published information, and use the ratio of the key negative published information to the published information of the key publisher as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

[0026] A further technical solution is to perform word segmentation processing on the published information of the publisher entity excluding the key publishers to obtain common word segments, and match the common word segments with the keyword vocabulary to obtain the number of common negative words.

[0027] A further technical solution is to use the number of common negative words as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

[0028] A further technical solution is to perform weighted processing on the number of negative English word segmentations, the number of negative Pinyin word segmentations, the number of negative Pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations of the key publishers, and use the weighted number of negative English word segmentations, the number of negative Pinyin word segmentations, the number of negative Pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations of the key publishers as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

[0029] Since information published with typos or abbreviations often contains more negative emotions, the accuracy and specificity of the prediction are further improved by weighting the above numbers.

[0030] On the other hand, the present invention provides a network public opinion information monitoring and processing system, which adopts the above-mentioned network public opinion information monitoring and processing method, specifically comprising:

[0031] Information extraction module, information matching module, model processing module, result output module;

[0032] The information extraction module is responsible for obtaining the network public opinion information to be monitored, extracting the publisher of the network public opinion information, and building a monitor database based on the historical publishing information of the account subject of the monitoring platform of the network public opinion information;

[0033] The information matching module is responsible for matching the publisher subject with the monitor database, and obtaining the key publisher subject according to the matching result; extracting the published information of the key publisher subject, and performing segmentation processing on the published information to obtain basic segmentations; matching the basic segmentations with the keyword thesaurus to obtain the number of basic negative segmentations; performing English translation based on the segmentations except the basic negative words to form English segmentations, and comparing the English segmentations with the English keyword database to determine the number of negative English segmentations, based on excluding negative English segmentations and basic negative words, and matching the pinyin, pinyin initials, and English initials of the segmentations excluding negative English segmentations and basic negative words with the pinyin keyword database, the pinyin initials database, and the English initials database, respectively, to obtain the number of negative pinyin segmentations, the number of negative pinyin initials segmentations, and the number of English initials segmentations;

[0034] The model processing module is responsible for feeding the proportion of the key publishers in the publishers, the number of basic negative segmented words, the number of negative English segmented words, the number of negative pinyin segmented words, the number of negative pinyin abbreviation segmented words, and the number of English abbreviation segmented words into the public opinion prediction model based on the EPO-LSTM algorithm;

[0035] The result output module is responsible for outputting the public opinion judgment results.

[0036] Other features and advantages will be described in the following description, and partly become apparent from the description, or understood by practicing the invention. The purpose and other advantages of the invention are realized and obtained by the structures particularly pointed out in the description and the drawings.

[0037] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings.

[0039] Figure 1 It is a flow chart of a network public opinion information monitoring and processing method according to Example 1.

[0040] Figure 2 It is a flowchart of the specific steps of constructing the monitor database according to Example 1.

[0041] Figure 3 This is a structural diagram of a network public opinion information monitoring and processing system according to Example 2. DETAILED DESCRIPTION

[0042] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that the present invention will be comprehensive and complete and fully convey the concepts of the example embodiments to those skilled in the art. The same reference numerals in the figures represent the same or similar structures, and thus their detailed description will be omitted.

[0043] The terms "a", "an", "the", and "said" are used to indicate the presence of one or more elements / components / etc.; the terms "comprising" and "having" are used to express an open-ended inclusive meaning and mean that additional elements / components / etc. may be present in addition to the listed elements / components / etc.

[0044] Today, as network technology matures, the gradual development and popularization of Internet technology, especially mobile Internet technology, has integrated the Internet into all aspects of social life. With the lowering of the threshold for online social networking, netizens who previously could only search and obtain the information they needed from the Internet can now participate in the exchange and creation of online information by creating their own social media or logging into numerous online social platforms. A research report by the China Internet Information Center pointed out that as of June this year, the number of Chinese netizens has exceeded 1 billion.

[0045] At the same time, due to the existence of intentional spelling errors or a large number of abbreviated expressions on the Internet, inaccurate results may occur during the identification of online public opinion information. In order to solve the above problems, China's authorized invention patent publication number CN103544242B "A sentiment entity search system for Weibo" uses a constructed word relationship graph to expand queries and realize the identification of typos or abbreviated expressions. However, similar spelling errors or a large number of abbreviated expressions are often updated very quickly on the Internet, and there are more negative emotional components in public opinion information with similar spelling errors. Therefore, if targeted identification and expansion cannot be performed according to the common types of spelling errors, and supplementary verification is performed based on the publisher's historical information release, a large recognition error rate will be caused, and the results of online public opinion monitoring will be inaccurate, so that the public opinion situation cannot be paid attention to in the first time, and once a risk occurs, it will bring serious consequences.

[0046] Example 1

[0047] To solve the above problems, according to one aspect of the present invention, Figure 1 As shown, a method for monitoring and processing network public opinion information is provided.

[0048] A method for monitoring and processing network public opinion information, characterized in that it specifically includes:

[0049] S11 obtains the network public opinion information to be monitored, extracts the publisher of the network public opinion information, and constructs a monitor database based on the historical publishing information of the account subject of the monitoring platform of the network public opinion information;

[0050] S12: matching the publisher subject with the monitor database, and obtaining a key publisher subject based on the matching result;

[0051] S13 extracts the published information of the key publisher, and performs word segmentation processing on the published information to obtain basic word segmentation;

[0052] S14 matches the basic segmentation words with the keyword word library to obtain the number of basic negative segmentation words; performs English translation based on the segmentation words other than the basic negative words to form English segmentation words, and compares the English segmentation words with the English keyword database to determine the number of negative English segmentation words, based on the segmentation words excluding negative English segmentation words and basic negative words, and matches the pinyin, pinyin abbreviation, and English abbreviation of the segmentation words excluding negative English segmentation words and basic negative words with the pinyin keyword database, the pinyin abbreviation database, and the English abbreviation database, respectively, to obtain the number of negative pinyin segmentation words, the number of negative pinyin abbreviation segmentation words, and the number of English abbreviation segmentation words;

[0053] S15 is based on the proportion of the key publishers among the publishers, the number of basic negative word segmentations, the number of negative English word segmentations, the number of negative pinyin word segmentations, the number of negative pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations, and inputs them into the public opinion prediction model based on the EPO-LSTM algorithm to output the public opinion judgment result.

[0054] First, the network public opinion information to be monitored is obtained, and the publisher subject is extracted, and a monitor database is constructed based on the historical release information, and the publisher subject is matched with the monitor database to find the key publisher subject. Since the key publisher is more likely to use typos or abbreviations, the key publishers are first screened, and based on the screening results, the release information of the key publishers is segmented, and the obtained segmented words are matched with the keyword thesaurus to obtain the number of basic negative segmented words, and the segmented words excluding the basic negative words are translated into English to form English segmented words, and the English segmented words are compared with the English keyword database to determine the number of negative English segmented words, based on the segmented words excluding negative English segmented words and basic negative words, and the pinyin, pinyin initials abbreviation, and English initials abbreviation of the segmented words excluding negative English segmented words and basic negative words are matched with the pinyin keyword database, the pinyin initials abbreviation database, and the English initials abbreviation database, respectively. , and obtain the number of negative pinyin segmentations, the number of negative pinyin abbreviation segmentations, and the number of English abbreviation segmentations, thereby solving the original inaccurate recognition problem caused by similar spelling errors or a large number of abbreviated expressions on the Internet, the update speed is very fast, and there are more negative emotional components in the public opinion information with similar spelling errors. By performing targeted recognition and expansion according to the common types of spelling errors, and supplementing verification based on the publisher's historical release information, the recognition result and recognition efficiency are greatly improved, further laying the foundation for realizing accurate judgment of public opinion information in the first time, and based on the proportion of the key publisher subject in the publisher, the basic negative segmentation number, the negative English segmentation number, the negative pinyin segmentation number, the negative pinyin abbreviation segmentation number, and the English abbreviation segmentation number are sent to the public opinion prediction model based on the EPO-LSTM algorithm to output the public opinion judgment result.

[0055] By establishing a monitor database based on historical information, a foundation is laid for further realizing the identification of key publishers, and further based on the matching results of the monitor database and the publisher, the key publishers are obtained, and the screening of people who often use typos and abbreviations is realized, thereby further improving the overall efficiency and accuracy. By matching the word segmentation results of the key publishers with the keyword database, the English keyword database, the pinyin keyword database, the pinyin abbreviation database, and the English abbreviation database, the basic negative word segmentation quantity is obtained, the negative English word segmentation quantity, the negative pinyin word segmentation quantity, the negative pinyin abbreviation word segmentation quantity, and the English abbreviation word segmentation quantity, thereby realizing the word segmentation quantity of negative words. The quantitative classification type matching further improves the accuracy of the recognition results and ensures the overall efficiency. The public opinion prediction model based on the EPO-LSTM algorithm is input based on the proportion of the key publishers, the number of basic negative word segmentations, the number of negative English word segmentations, the number of negative pinyin word segmentations, the number of negative pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations to output the public opinion judgment results, thereby solving the problem of the original statistical results based on all the people, which leads to low accuracy and lack of pertinence. It can achieve accurate judgment of public opinion, and take corresponding measures according to the results of public opinion judgment. The output public opinion judgment results are divided into five stages: germination period, rising period, outbreak period, stable period, and decline period.

[0056] In another possible embodiment, the word segmentation processing method adopts the Jieba word segmentation model.

[0057] In another possible embodiment, Figure 2 As shown, the specific steps of constructing the monitor database are:

[0058] S11: performing word segmentation processing on the historical release information of the account subject to obtain historical word segmentation, and matching the historical word segmentation with the keyword database to obtain negative historical word segmentation;

[0059] S12 obtains historical proportion data based on the proportion of the release information of the negative historical segmentation words in the historical distribution information;

[0060] S13: When the historical proportion data is greater than a first proportion threshold, the account subject is determined to be a monitor, and the monitor is placed in the monitor database.

[0061] For a specific example, when the historical proportion data is 70% and the first proportion threshold is 60%, the account subject is determined to be a monitor.

[0062] By using the historical proportion data of the release information of negative historical segmentation words in the historical distribution information, it is determined whether the account subject is a monitor, thereby further achieving accurate positioning and analysis of the monitor. When the historical proportion data is greater than the first proportion threshold, it means that there is a lot of negative release information at this time, so that the monitor can be accurately located. At the same time, it also solves the one-sidedness of the original judgment based solely on quantity or words, so that the accuracy is enhanced to a certain extent.

[0063] In another possible embodiment, the first percentage threshold is determined according to the number of account entities of the monitoring platform and the number of historical published information.

[0064] For example, the number of account entities on the monitoring platform is 100, which is a small number. The first percentage threshold can be set to 20%. When the number of historical published information of the account entity is 10, which is not a large number, the first percentage threshold can be set to 90%. In other words, only when 9 out of 10 historical published information have problems will the account entity be listed as a monitor.

[0065] When the number of account entities is small and less than the first account threshold, the first proportion threshold can be set larger. When the number of account entities is greater than or equal to the first account threshold, it means that the number is large at this time, and the first proportion threshold can be lowered. When the number of historical published information is small and less than the first information threshold, the first proportion threshold can be set larger. When the number of historical published information is greater than or equal to the first information threshold, it means that the number is large at this time, and the first proportion threshold can be lowered. The setting of the first proportion threshold can be further realized according to the actual situation, so that the setting result becomes more reliable and applicable.

[0066] In another possible embodiment, when the number of historical published information is less than a first quantity threshold and the historical proportion data is greater than a second proportion threshold, the account subject is determined to be a monitor; when the number of historical published information is greater than the second quantity threshold and the historical proportion data is greater than a third proportion threshold, the account subject is determined to be a monitor, the third proportion threshold is less than the first proportion threshold, the first proportion threshold is less than the second proportion threshold, and the first quantity threshold is less than the second quantity threshold.

[0067] To give a specific example, when the first quantity threshold is 5, the second proportion threshold is 90%, that is to say, for account entities with less than 5 items, only those with a proportion of more than 90% will be listed as monitors. When the second quantity threshold is 200, the third proportion threshold can be set to 50%, and only those with a proportion of more than 50% will be listed as monitors.

[0068] The judgment of the monitor is made based on the number of historically published information of the account subject, thus solving the problem of misjudgment of account subjects with a small or large number of published information due to the use of a unified standard regardless of the number of historical published information, and further improving the judgment.

[0069] In another possible embodiment, the published information containing the basic negative words, the published information containing the number of negative English word segmentations, the published information containing the number of negative pinyin word segmentations, the published information containing the number of negative pinyin abbreviation word segmentations, and the published information containing the number of English abbreviation word segmentations are taken as key negative published information, and the ratio of the key negative published information to the published information of the key publisher entity is taken as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

[0070] In another possible embodiment, the published information of the publisher entity excluding the key publishers is segmented to obtain common segmented words, and the common segmented words are matched with the keyword thesaurus to obtain the number of common negative words.

[0071] In another possible embodiment, the number of common negative words is used as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

[0072] In another possible embodiment, the number of negative English word segmentations, the number of negative Pinyin word segmentations, the number of negative Pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations of the key publishers are weighted, and the weighted number of negative English word segmentations, the number of negative Pinyin word segmentations, the number of negative Pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations of the key publishers are used as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

[0073] To take a specific example, if the weight of the number of basic segmentations is 1, then the weights of the number of negative English segmentations, the number of negative Pinyin segmentations, the number of negative Pinyin abbreviation segmentations, and the number of English abbreviation segmentations of the key publishers can all be constants greater than 1, or the number of negative English segmentations, the number of negative Pinyin segmentations, the number of negative Pinyin abbreviation segmentations, and the number of English abbreviation segmentations of the key publishers can be arranged from large to small according to the weights of the number of negative Pinyin segmentations, the number of negative English segmentations, the number of negative Pinyin abbreviation segmentations, and the number of English abbreviation segmentations. Since the subjective negative emotions of using negative Pinyin and negative English are more, a larger weight is given, thereby further improving the accuracy and precision of the overall judgment.

[0074] Since information published with typos or abbreviations often contains more negative emotions, the accuracy and specificity of the prediction are further improved by weighting the above numbers.

[0075] Example 2

[0076] like Figure 3 As shown, the present invention provides a network public opinion information monitoring and processing system, which adopts the above-mentioned network public opinion information monitoring and processing method, specifically comprising:

[0077] Information extraction module, information matching module, model processing module, result output module;

[0078] The information extraction module is responsible for obtaining the network public opinion information to be monitored, extracting the publisher of the network public opinion information, and building a monitor database based on the historical publishing information of the account subject of the monitoring platform of the network public opinion information;

[0079] The information matching module is responsible for matching the publisher subject with the monitor database, and obtaining the key publisher subject according to the matching result; extracting the published information of the key publisher subject, and performing segmentation processing on the published information to obtain basic segmentations; matching the basic segmentations with the keyword thesaurus to obtain the number of basic negative segmentations; performing English translation based on the segmentations except the basic negative words to form English segmentations, and comparing the English segmentations with the English keyword database to determine the number of negative English segmentations, based on excluding negative English segmentations and basic negative words, and matching the pinyin, pinyin initials, and English initials of the segmentations excluding negative English segmentations and basic negative words with the pinyin keyword database, the pinyin initials database, and the English initials database, respectively, to obtain the number of negative pinyin segmentations, the number of negative pinyin initials segmentations, and the number of English initials segmentations;

[0080] The model processing module is responsible for feeding the proportion of the key publishers in the publishers, the number of basic negative segmented words, the number of negative English segmented words, the number of negative pinyin segmented words, the number of negative pinyin abbreviation segmented words, and the number of English abbreviation segmented words into the public opinion prediction model based on the EPO-LSTM algorithm;

[0081] The result output module is responsible for outputting the public opinion judgment results.

[0082] In the embodiments of the present invention, the term "plurality" refers to two or more than two, unless otherwise clearly defined. The terms "installation", "connection", "fixation" and the like should be understood in a broad sense. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection. For those skilled in the art, the specific meanings of the above terms in the embodiments of the present invention can be understood according to the specific circumstances.

[0083] In the description of the embodiments of the present invention, it needs to be understood that the directions or positional relationships indicated by the terms "upper" and "lower" are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the embodiments of the present invention and simplifying the description, rather than indicating or implying that the device or unit referred to must have a specific direction, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the embodiments of the present invention.

[0084] In the description of this specification, the description of the terms "one embodiment", "a preferred embodiment", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0085] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for monitoring and processing network public opinion information, characterized in that: Specifically include: S11 obtains the network public opinion information to be monitored, extracts the publisher of the network public opinion information, and constructs a monitor database based on the historical publishing information of the account subject of the monitoring platform of the network public opinion information; S12: matching the publisher subject with the monitor database, and obtaining a key publisher subject based on the matching result; S13 extracts the published information of the key publisher, and performs word segmentation processing on the published information to obtain basic word segmentation; S14 matches the basic segmentation words with the keyword word library to obtain the number of basic negative segmentation words; performs English translation based on the segmentation words other than the basic negative words to form English segmentation words, and compares the English segmentation words with the English keyword database to determine the number of negative English segmentation words, based on the segmentation words excluding negative English segmentation words and basic negative words, and matches the pinyin, pinyin initials, and English initials of the segmentation words excluding negative English segmentation words and basic negative words with the pinyin keyword database, the pinyin initials database, and the English initials database, respectively, to obtain the number of negative pinyin segmentation words, the number of negative pinyin initials segmentation words, and the number of English initials segmentation words; S15 sends the proportion of the key publisher entities among the publishers, the number of basic negative word segmentations, the number of negative English word segmentations, the number of negative pinyin word segmentations, the number of negative pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations to the public opinion prediction model based on the EPO-LSTM algorithm, and outputs the public opinion judgment result.

2. A method for monitoring and processing network public opinion information according to claim 1, characterized in that: The word segmentation processing method adopts the jieba word segmentation model.

3. A method for monitoring and processing network public opinion information according to claim 1, characterized in that: The specific steps of constructing the monitor database are: S11: performing word segmentation processing on the historical release information of the account subject to obtain historical word segmentation, and matching the historical word segmentation with the keyword database to obtain negative historical word segmentation; S12 obtaining historical proportion data based on the proportion of the published information of the negative historical segmentation words in the historical distribution information; S13: When the historical proportion data is greater than a first proportion threshold, the account subject is determined to be a monitor, and the monitor is placed in the monitor database.

4. A method for monitoring and processing network public opinion information according to claim 3, characterized in that: The first percentage threshold is determined according to the number of account entities of the monitoring platform and the number of historical published information.

5. A method for monitoring and processing network public opinion information according to claim 3, characterized in that: When the number of historical published information is less than the first quantity threshold and the historical proportion data is greater than the second proportion threshold, the account subject is determined to be a monitor; when the number of historical published information is greater than the second quantity threshold and the historical proportion data is greater than the third proportion threshold, the account subject is determined to be a monitor, the third proportion threshold is less than the first proportion threshold, the first proportion threshold is less than the second proportion threshold, and the first quantity threshold is less than the second quantity threshold.

6. A method for monitoring and processing network public opinion information according to claim 1, characterized in that: The published information where the basic negative words are located, the published information where the negative English word segmentation is located, the published information where the negative pinyin word segmentation is located, the published information where the negative pinyin abbreviation word segmentation is located, and the published information where the English abbreviation word segmentation is located is the key negative published information, and the ratio of the key negative published information to the published information of the key publisher entity is used as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

7. A method for monitoring and processing network public opinion information according to claim 1, characterized in that: The published information of the publisher subject excluding the key publisher is segmented to obtain common segmented words, and the common segmented words are matched with the keyword word library to obtain the number of common negative words.

8. A method for monitoring and processing network public opinion information according to claim 7, characterized in that: The number of common negative words is used as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

9. A method for monitoring and processing network public opinion information according to claim 1, characterized in that: The number of negative English word segmentations, the number of negative Pinyin word segmentations, the number of negative Pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations of the key publishers are weighted, and the weighted number of negative English word segmentations, the number of negative Pinyin word segmentations, the number of negative Pinyin abbreviation word segmentations, and the number of English abbreviation word segmentations of the key publishers are used as one of the inputs of the public opinion prediction model to predict the public opinion judgment result.

10. A network public opinion information monitoring and processing system, using a network public opinion information monitoring and processing method according to any one of claims 1 to 9, specifically comprising: Information extraction module, information matching module, model processing module, result output module; The information extraction module is responsible for obtaining the network public opinion information to be monitored, extracting the publisher of the network public opinion information, and building a monitor database based on the historical publishing information of the account subject of the monitoring platform of the network public opinion information; The information matching module is responsible for matching the publisher subject with the monitor database, and obtaining the key publisher subject according to the matching result; extracting the published information of the key publisher subject, and performing segmentation processing on the published information to obtain basic segmentations; matching the basic segmentations with the keyword thesaurus to obtain the number of basic negative segmentations; performing English translation based on the segmentations except the basic negative words to form English segmentations, and comparing the English segmentations with the English keyword database to determine the number of negative English segmentations, based on excluding negative English segmentations and basic negative words, and matching the pinyin, pinyin initials, and English initials of the segmentations excluding negative English segmentations and basic negative words with the pinyin keyword database, the pinyin initials database, and the English initials database, respectively, to obtain the number of negative pinyin segmentations, the number of negative pinyin initials segmentations, and the number of English initials segmentations; The model processing module is responsible for feeding the proportion of the key publishers in the publishers, the number of basic negative segmented words, the number of negative English segmented words, the number of negative pinyin segmented words, the number of negative pinyin abbreviation segmented words, and the number of English abbreviation segmented words into the public opinion prediction model based on the EPO-LSTM algorithm; The result output module is responsible for outputting the public opinion judgment results.

Images