A public opinion intelligent identification system and method

Abstract

The present invention belongs to the technical field of public opinion identification, and specifically discloses a public opinion intelligent identification system, including a Kafka data receiving module, a public opinion intelligent analysis module, a Kafka data push module and a form making module; the Kafka data receiving module is used to receive the public opinion data in the Kafka data stream; the public opinion intelligent analysis module is used to determine whether the public opinion data is real public opinion, and if it is real public opinion, the public opinion data is divided into categories and levels; the Kafka data push module is used to pass the public opinion identification results to the form making module in the order of the Kafka message queue; the form making module is used to generate a work order according to the public opinion data received from the Kafka data receiving module and the public opinion identification results received by the Kafka data push module. The present invention judges and identifies the public opinion data through the public opinion intelligent analysis module, which can reduce the work intensity of public opinion researchers and judges; and can avoid inaccurate classification caused by manual classification.

Description

Technical Field

[0001] The present invention belongs to the technical field of public opinion recognition, and in particular to a public opinion intelligent recognition system. Background Art

[0002] As the State Grid Corporation of China pays more attention to public opinion, the power supply service public opinion has the disadvantages of high service risk, decentralized control affecting the overall service level, high operating costs, and low service efficiency. In order to improve the State Grid Corporation of China's ability to prevent and respond to power supply service incidents, to prevent and reduce the losses and impacts caused by power supply service incidents to the greatest extent, and to maintain the company's brand image, the company decided to improve the public opinion monitoring system.

[0003] At present, the public opinion information of the public opinion monitoring system needs to be initially judged by the dispatcher. After the judgment, the public opinion work order is sent to the corresponding customer service department. The customer service specialist will once again identify in detail whether it belongs to public opinion, the type of public opinion, the unit of public opinion information, etc. Since there are tens of thousands of public opinion data every day, manual repeated judgment is caused, and the workload is large. The accuracy of the judgment is greatly affected by the judgment of the supervisor. In order to reduce the workload of public opinion judgment personnel and customer service specialists, improve the accuracy of public opinion recognition, and realize the automatic real-time public opinion monitoring function, it is urgent to propose a public opinion intelligent recognition system.

[0004] There is currently no solution to the shortcomings of the existing technology. Summary of the invention

[0005] Purpose of the invention: To provide a public opinion intelligent identification system to solve the above-mentioned problems existing in the prior art.

[0006] Technical solution: A public opinion intelligent identification system, including a Kafka data receiving module, a public opinion intelligent analysis module, a Kafka data push module and a bill making module; wherein:

[0007] The Kafka data receiving module is used to receive the public opinion data in the Kafka data stream and send the public opinion data to the public opinion intelligent analysis module and the order making module;

[0008] The public opinion intelligent analysis module is used to determine whether the public opinion data is real public opinion. If it is real public opinion, the public opinion data is divided into categories and levels to obtain public opinion identification results;

[0009] The Kafka data push module is used to pass the public opinion identification results to the order making module in the order of the Kafka message queue, so that the downstream system can generate work orders;

[0010] The order making module is used to generate work orders based on the public opinion data received from the Kafka data receiving module and the public opinion identification results received from the Kafka data pushing module.

[0011] Further: the public opinion intelligent analysis module includes a receiving submodule, a public opinion judgment module and a judgment and recognition module, wherein:

[0012] The receiving submodule is used to receive the public opinion data sent by the Kafka data receiving module and pass the public opinion data to the public opinion judgment module;

[0013] The public opinion judgment module is used to judge whether the public opinion corresponding to the public opinion data is real public opinion;

[0014] The judgment and identification module is used to judge the type of public opinion and the level of public opinion events.

[0015] Through this technical solution, the judgment and identification of public opinion data can be realized, thereby reducing the workload and work intensity of public opinion analysts.

[0016] Further: the public opinion judgment module includes a judgment submodule and a storage submodule;

[0017] The judgment submodule is used to judge whether the public opinion data is real public opinion. If it is real public opinion, the public opinion data is passed to the judgment and identification module. If it is not real public opinion, the data is saved in the saving submodule.

[0018] Through this technical solution, data that is not real public opinion can be saved in the storage submodule, thereby reducing the public opinion data flowing into the judgment and identification module and reducing the memory occupancy of the judgment and identification module due to judging non-real public opinion.

[0019] Further: the judgment and identification module specifically includes a level judgment module and a type judgment module, wherein:

[0020] The level judgment module is used to identify the level of public opinion events, which include general, major and serious levels;

[0021] The type judgment module is used to compare the public opinion data with the preset public opinion types to determine the type of public opinion. The preset public opinion types include power supply service public opinion, State Grid customer service center public opinion and positive publicity and reporting public opinion.

[0022] Through this technical solution, it is possible to realize the level judgment and type identification of public opinion. The type identification of public opinion can facilitate the distribution of public opinion to the corresponding units downstream, and the identification of the level of public opinion can facilitate the processing of public opinion according to its level.

[0023] In addition, the present invention also provides a method for intelligently identifying public opinion, comprising the following steps:

[0024] S1: The Kafka data receiving module receives the public opinion text data and transmits the public opinion text data to the public opinion intelligent analysis module and the order making module;

[0025] S2: The public opinion intelligent analysis module determines whether the received public opinion text data is public opinion data. If it is public opinion data, it determines the type of public opinion data and the level of the public opinion event, and passes the category and level corresponding to the public opinion text data to the Kafkka data push module;

[0026] S3: The Kafka data push module transmits the data to the order making module in the order of the Kafka message queue;

[0027] S4: The order making module generates a work order based on the public opinion data received from the Kafka data receiving module and the public opinion identification results received from the Kafka data pushing module.

[0028] Further: Step S2 specifically includes:

[0029] S2.1: The receiving submodule receives the public opinion data sent by the Kafka data receiving module, and transmits the public opinion data to the public opinion judgment module;

[0030] S2.2: The judgment submodule of the public opinion judgment module judges whether the received public opinion data is real public opinion. If it is real public opinion, the public opinion data is transmitted to the judgment and identification module. If it is not real public opinion, the data is saved to the saving submodule of the public opinion judgment module.

[0031] S2.3: After the judgment and identification module receives the public opinion data of real public opinion, it judges the type of public opinion and the level of the public opinion event based on the public opinion data, and passes the judgment result to the Kafka data push module. The public opinion types include power supply service public opinion, State Grid Customer Service Center public opinion and positive publicity and reporting public opinion. The levels of public opinion events include general, major and serious.

[0032] Through this technical solution, it is possible to judge the public opinion data and classify the real public opinion data into levels and types, so as to facilitate the transmission of the real public opinion data to the corresponding downstream units.

[0033] Further: Step S4 specifically includes:

[0034] If the order-making module receives the public opinion text data transmitted by the Kafka data receiving module but does not receive the public opinion judgment result pushed by the Kafka data pushing module, the order-making module will generate a work order for the public opinion data after a delay of half an hour according to the delay mechanism.

[0035] Through this technical solution, a half-hour delay mechanism is set up, which can basically analyze and judge the public opinion data, and avoid the impact of too long a delay on the subsequent processing of the public opinion data.

[0036] Further: if the order making module receives the public opinion judgment result pushed by the Kafka data push module within the half-hour delay time, the order making module will automatically make an order, that is, make the public opinion data into a work order according to the public opinion judgment result and automatically pass the work order to the corresponding downstream receiving unit;

[0037] If the order-making module does not receive the public opinion judgment result pushed by the Kafka data push module within the half-hour delay time, the order-making module will semi-automatically create a work order, which will be manually distributed to the downstream receiving unit;

[0038] The corresponding downstream receiving units include the power supply service receiving unit, the State Grid Customer Service Center receiving unit and the positive publicity and reporting receiving unit.

[0039] The present invention discloses a public opinion intelligent identification system, which has the following beneficial effects:

[0040] 1. The present invention uses the public opinion intelligent analysis module to judge and identify public opinion data, which can reduce the workload of public opinion researchers;

[0041] 2. The present invention can convert public opinion data into work orders according to the type of public opinion and transmit them to the corresponding downstream processing units, thereby avoiding inaccurate classification caused by manual classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a module block diagram of a public opinion intelligent identification system of the present invention;

[0043] Figure 2 It is a method flow chart of a method for intelligent identification of public opinion of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0045] like Figure 1 As shown, a public opinion intelligent identification system includes a Kafka data receiving module, a public opinion intelligent analysis module, a Kafka data push module and a bill making module; wherein:

[0046] The Kafka data receiving module is used to receive public opinion data in the Kafka data stream, and convert the public opinion data into JSON format and send it to the public opinion intelligent analysis module and the order making module;

[0047] The public opinion intelligent analysis module is used to determine whether the public opinion data is real public opinion. If it is real public opinion, the public opinion data is divided into categories and levels to obtain public opinion identification results, and the public opinion identification results are pushed into the Kafka data push module in json format;

[0048] The Kafka data push module is used to pass the public opinion identification results to the order making module in the order of the Kafka message queue, so that the downstream system can generate work orders;

[0049] The order making module is used to generate work orders based on the public opinion data received from the Kafka data receiving module and the public opinion identification results received from the Kafka data pushing module.

[0050] In the above embodiment, after the Kafka data receiving module receives the public opinion data stream, it sends the public opinion data stream in json format to the public opinion intelligent analysis module and the order making module. The public opinion intelligent analysis module is used to screen the public opinion data, determine whether the public opinion data is real public opinion data, and whether the public opinion meets the public opinion range. If it is real public opinion data, the level and type of the public opinion data are divided to obtain the public opinion recognition result and push it into the Kafka data push module in json format. The Kafka data push module then passes the public opinion recognition result to the order making module. The analysis module needs to identify and judge public opinion. Therefore, for the same public opinion data, the order-making module first receives the public opinion data sent by the Kafka data receiving module, and then receives the identification result of the public opinion. Therefore, the order-making module is designed with a delay mechanism. That is, if the order-making module receives the identification result of the public opinion within half an hour, the public opinion identification result and the public opinion data will be made into a work order and passed to the corresponding downstream processing unit. If the order-making module does not receive the identification result of the public opinion within half an hour, the order-making module will make the public opinion data into a work order, and the work order will be manually assigned to the downstream processing unit.

[0051] In this embodiment: the public opinion intelligent analysis module includes a receiving submodule, a public opinion judgment module and a judgment identification module, wherein: the receiving submodule is used to receive the public opinion data sent by the Kafka data receiving module, and pass the public opinion data to the public opinion judgment module; the public opinion judgment module is used to judge whether the public opinion corresponding to the public opinion data is real public opinion; the judgment identification module is used to judge the type of public opinion and the level of public opinion events; in this embodiment, after the receiving submodule receives the public opinion data from the Kafka data receiving module, the public opinion judgment module passes the public opinion data to the public opinion judgment module, wherein the public opinion judgment module includes a judgment submodule and a saving submodule, and the judgment submodule judges whether the received public opinion data is real public opinion. The judgment here is made by judging the public opinion data from the text results, vocabulary level and sentence level through a stacked recurrent neural network and an attention mechanism. If it is real public opinion, the judgment identification module identifies the type and level of public opinion. If it is not real public opinion, the data is saved in the saving submodule; the judgment identification module includes a level judgment module and a type judgment module, wherein the type of public opinion is determined by the type judgment module. The public opinion data is obtained by comparing it with the preset public opinion types, which include power supply service public opinion, State Grid customer service center public opinion and positive publicity and reporting public opinion. By comparing the public opinion data with the preset public opinion types, it can be obtained that the public opinion data belongs to the power supply service public opinion type or the State Grid customer service center public opinion or the positive publicity and reporting public opinion, so that the public opinion data can be sent to the order making module through the Kafka data push module according to the type; the level of the public opinion event is realized by the level judgment module. After the level judgment module extracts the keywords of the public opinion data, it judges whether the public opinion data is a general, large and major event according to the proportion of the keywords in the public opinion database. For example, if the keyword of the public opinion data is cost, the level judgment module identifies the proportion of the public opinion data containing the keyword in all the public opinion data in the public opinion database, and then obtains the level of the public opinion data according to the proportion. In the design, the public opinion data with a proportion of 30% is classified as a general level event, the public opinion data with a proportion between 30% and 80% is classified as a large level event, and the public opinion data with a proportion between 80% and 100% is classified as a major public opinion event.

[0052] The identification process of the type judgment module is as follows: first, the words in the sentences in the public opinion data are encoded and converted into word vectors, and then input into the bidirectional GRU neural network, and the hidden state output corresponding to the word is obtained by combining the context information, and the vectors of the words in the sentence are weighted summed using the Attention mechanism to obtain the vector representation of each sentence; according to the sentence vector, the bidirectional GRU is used to construct the document vector and the document vector is used to obtain the document representation using the Attention mechanism, and finally a fully connected soft max layer is connected for multi-label classification to obtain the label classification of the public opinion data, and then the label classification is compared with the preset public opinion type to obtain the public opinion data belongs to the power supply service public opinion type or the State Grid Customer Service Center public opinion or the positive publicity report public opinion.

[0053] In addition, the present invention also provides a method for intelligent identification of public opinion. Figure 2 As shown, the following steps are included:

[0054] S1: The Kafka data receiving module receives the public opinion text data and transmits the public opinion text data to the public opinion intelligent analysis module and the order making module;

[0055] S2: The public opinion intelligent analysis module determines whether the received public opinion text data is public opinion data. If it is public opinion data, it determines the type of public opinion data and the level of the public opinion event, and transmits the category and level corresponding to the public opinion text data to the Kafka data push module; this step specifically includes:

[0056] S2.1: The receiving submodule receives the public opinion data sent by the Kafka data receiving module, and transmits the public opinion data to the public opinion judgment module;

[0057] S2.2: The judgment submodule of the public opinion judgment module judges whether the received public opinion data is real public opinion. If it is real public opinion, the public opinion data is transmitted to the judgment and identification module. If it is not real public opinion, the data is saved to the saving submodule of the public opinion judgment module.

[0058] S2.3: After the judgment and identification module receives the public opinion data of the real public opinion, it judges the type of public opinion and the level of the public opinion event according to the public opinion data, and transmits the judgment result to the Kafka data push module. The public opinion types include power supply service public opinion, State Grid customer service center public opinion and positive publicity and reporting public opinion. The levels of public opinion events include general, major and serious.

[0059] S3: The Kafka data push module transmits the data to the order making module in the order of the Kafka message queue;

[0060] S4: The order making module generates a work order based on the public opinion data received from the Kafka data receiving module and the public opinion identification result received by the Kafka data pushing module; this step specifically includes: if the order making module does not receive the public opinion judgment result pushed by the Kafka data pushing module after receiving the public opinion text data transmitted by the Kafka data receiving module, the order making module generates a work order from the public opinion data after a delay of half an hour according to the delay mechanism; that is, if the order making module receives the public opinion judgment result pushed by the Kafka data pushing module within the delay time of half an hour, the order making module automatically makes the order, that is, makes the public opinion data into a work order according to the public opinion judgment result and automatically transmits the work order to the corresponding downstream receiving unit;

[0061] If the order making module does not receive the public opinion judgment result pushed by the Kafka data push module within the half-hour delay time, the order making module semi-automatically makes the order, that is, the order making module generates a work order, which is manually assigned to the downstream receiving unit;

[0062] The corresponding downstream receiving units include the power supply service receiving unit, the State Grid Customer Service Center receiving unit and the positive publicity and reporting receiving unit.

[0063] In the above identification method, if the public opinion intelligent analysis identifies that the public opinion data belongs to the general level of public opinion in the power supply service, the order making module will make the public opinion data into a work order and send it to the power supply service receiving unit, and mark the public opinion as belonging to the general level.

[0064] In the above embodiment, before a piece of public opinion data is sent to the Kafka data receiving module, it will be marked with two group IDs (group ID 1 and group ID 2), where group ID 1 indicates that it is consumed by the order making module, and group ID2 is passed from the Kafka data receiving module to the public opinion intelligent analysis module for consumption, thereby obtaining the content ID in the public opinion intelligent analysis module; after the order making module obtains the public opinion data of group ID 1, it delays for one hour according to the delay mechanism. If the order making module obtains the content ID data of the public opinion intelligent analysis module within half an hour, the order making module automatically generates an order based on groupID 1 and content ID; if the order making module does not obtain the content ID data of the public opinion intelligent analysis module within half an hour, the order making module generates a work order, which is manually distributed to the downstream receiving unit.

[0065] The preferred embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical scheme of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A public opinion intelligent identification system, characterized in that: It includes Kafka data receiving module, public opinion intelligent analysis module, Kafka data push module and order making module; among them: The Kafka data receiving module is used to receive the public opinion data in the Kafka data stream and send the public opinion data to the public opinion intelligent analysis module and the order making module; The public opinion intelligent analysis module is used to determine whether the public opinion data is real public opinion. If it is real public opinion, the public opinion data is divided into categories and levels to obtain public opinion identification results; The Kafka data push module is used to pass the public opinion identification results to the order making module in the order of the Kafka message queue, so that the downstream system can generate work orders; The order making module is used to generate work orders based on the public opinion data received from the Kafka data receiving module and the public opinion identification results received from the Kafka data pushing module; If the ticket-making module receives the public opinion text data transmitted by the Kafka data receiving module but does not receive the public opinion judgment result pushed by the Kafka data pushing module, the ticket-making module will generate a work order for the public opinion data after a delay of half an hour according to the delay mechanism; If the order-making module receives the public opinion judgment result pushed by the Kafka data push module within the half-hour delay time, the order-making module will automatically generate a public opinion work order; If the order-making module does not receive the public opinion judgment result pushed by the Kafka data push module within the half-hour delay time, the order-making module will semi-automatically make an order. Semi-automatic order-making means that the order-making module generates a work order, which is manually assigned to the downstream receiving unit. The corresponding downstream receiving units include the power supply service receiving unit, the State Grid Customer Service Center receiving unit and the positive publicity and reporting receiving unit.

2. According to claim 1, the public opinion intelligent identification system is characterized by: The public opinion intelligent analysis module includes a receiving submodule, a public opinion judgment module and a judgment and recognition module, among which: The receiving submodule is used to receive the public opinion data sent by the Kafka data receiving module and pass the public opinion data to the public opinion judgment module; The public opinion judgment module is used to judge whether the public opinion corresponding to the public opinion data is real public opinion; The judgment and identification module is used to judge the type of public opinion and the level of public opinion events.

3. The public opinion intelligent identification system according to claim 2 is characterized by: The public opinion judgment module includes a judgment submodule and a storage submodule; The judgment submodule is used to judge whether the public opinion data is real public opinion. If it is real public opinion, the public opinion data is passed to the judgment and identification module. If it is not real public opinion, the data is saved in the saving submodule.

4. The public opinion intelligent identification system according to claim 2, characterized in that: The judgment and identification module specifically includes a level judgment module and a type judgment module, wherein: The level judgment module is used to identify the level of public opinion events, which include general, major and serious levels; The type judgment module is used to compare the public opinion data with the preset public opinion types to determine the type of public opinion. The preset public opinion types include power supply service public opinion, State Grid customer service center public opinion and positive publicity and reporting public opinion.

5. A method for intelligent identification of public opinion, characterized in that: The public opinion intelligent identification system according to any one of claims 1 to 4 is adopted, and specifically comprises the following steps: S1: The Kafka data receiving module receives the public opinion text data and transmits the public opinion text data to the public opinion intelligent analysis module and the order making module; S2: The public opinion intelligent analysis module determines whether the received public opinion text data is public opinion data. If it is public opinion data, it determines the type of public opinion data and the level of the public opinion event, and passes the category and level corresponding to the public opinion text data to the Kafkka data push module; S3: The Kafka data push module transmits the data to the order making module in the order of the Kafka message queue; S4: The order making module generates a work order based on the public opinion data received from the Kafka data receiving module and the public opinion identification results received from the Kafka data pushing module.

6. The method for intelligent identification of public opinion according to claim 5, characterized in that: Step S2 specifically includes: S2.1: The receiving submodule receives the public opinion data sent by the Kafka data receiving module, and transmits the public opinion data to the public opinion judgment module; S2.2: The judgment submodule of the public opinion judgment module judges whether the received public opinion data is real public opinion. If it is real public opinion, the public opinion data is transmitted to the judgment and identification module. If it is not real public opinion, the data is saved to the saving submodule of the public opinion judgment module. S2.3: After the judgment and identification module receives the public opinion data of real public opinion, it judges the type of public opinion and the level of the public opinion event based on the public opinion data, and passes the judgment result to the Kafka data push module. The public opinion types include power supply service public opinion, State Grid Customer Service Center public opinion and positive publicity and reporting public opinion. The levels of public opinion events include general, major and serious.

Images