A method and device for public opinion monitoring and conversation instant reconstruction for enterprise instant communication

Abstract

The application relates to a public opinion monitoring and conversation instant reconstruction method and device for enterprise instant communication, and belongs to the technical field of communication monitoring. Step one: a control system uniformly captures messages in customer chatting and group chatting, and carries out text extraction and arrangement on the captured messages, so as to complete the standardization pretreatment of the captured information; step two: the control system carries out sensitive word matching, generates a standardized public opinion triggering event record if a hit is obtained, and adds one to the number of public opinion hits, wherein the record at least comprises a member unique identifier, a target conversation identifier, a message ID, a conversation ID, a sending time, a conversation type, a sender, a hit keyword and a belonging group; step three: a routing strategy is executed on the public opinion triggering event, and an idempotent key is used for checking; when clicking is completed, step four is entered; step four: conversation reconstruction is carried out, the chatting message triggering the sensitive word is taken as a core, context backtracking loading is carried out, and the loading range is confirmed by the control system.

Description

Technical Field

[0001] This invention belongs to the field of communication monitoring technology, specifically relating to a method and apparatus for monitoring public opinion and reconstructing conversations in enterprise instant messaging. Background Technology

[0002] During internet chats, for security, confidentiality, or to maintain stability, it is necessary to monitor specific chats or intervene when necessary.

[0003] Traditional public opinion monitoring systems rely on manual monitoring of millions of data points daily to analyze online public opinion information. This approach significantly reduces the efficiency of online public opinion monitoring. Improvements, such as those using incomplete semantic feature vectors of pre-defined reference online public opinion information, inevitably exclude public opinion information outside the pre-defined reference information, hindering new public opinion discovery and monitoring. To address this, Chinese patent CN114385890B discloses an internet public opinion monitoring system, comprising: a data acquisition module for collecting and acquiring online public opinion information to be monitored; a communication unit or transmission interface for connecting to the data acquisition module and sending the collected online public opinion information to a processing device; the processing device includes: a comparison module, an analysis module, a machine learning system, a main classifier, an auxiliary classifier, a dynamic allocator, a configuration module, and a monitoring module. This invention can output keywords accumulated during the monitoring process to monitor the potential of these keywords to become trending topics.

[0004] However, the above-mentioned solutions are only general public opinion collection, analysis and output, and fail to be specifically optimized for the enterprise instant messaging scenario. In the enterprise instant messaging scenario, conversations are often both immediate and private. The immediacy makes it necessary to monitor and handle public opinion quickly, while the privacy, that is, normal daily conversations, need to be conducted privately and cannot be disclosed, which makes it impossible to reconstruct and display the monitored conversations on the handler's interface in real time. Therefore, there is a need for an enterprise instant messaging public opinion monitoring and conversation real-time reconstruction method and device that can quickly handle and reconstruct conversations. Summary of the Invention

[0005] To address the aforementioned problems in the existing technology, this invention provides a method and apparatus for monitoring public opinion and reconstructing conversations in enterprise instant messaging, which features rapid handling and rapid conversation reconstruction.

[0006] The objective of this invention can be achieved through the following technical solutions: A method for monitoring public opinion and reconstructing conversations in real-time for enterprise instant messaging includes the following steps: Step 1: The control system uniformly captures messages from chats with customers and group chats, and extracts and organizes the captured messages into text, completing the standardized preprocessing of the captured information. Step 2: Control the system to perform sensitive word matching; match the text sequence with the sensitive word group and the dictionary. If a match is found, generate a standardized public opinion trigger event record and increment the public opinion hit count. The record should include at least: member unique identifier, target session identifier, message ID, session ID, sending time, session type, sender, hit keywords and the group to which they belong.

[0007] Step 3: Execute routing strategy for public opinion triggering events: Determine if the scenario is a group chat. If the result is yes, randomly select a member of the enterprise as the first person to handle the situation. At the same time, use idempotent keys to verify that only one valid reminder is triggered. Then, for the first person to handle the situation, create an entry point for the session where the public opinion triggering event is located in the workbench and wait for the first person to handle the situation to click. After the click is completed, proceed to step 4. Step 4: Reconstruct the session by focusing on the chat message that triggered the sensitive word and performing context backloading. The loading scope is determined by the control system.

[0008] As a preferred technical solution of the present invention, step four further includes: controlling the system to count the number of sensitive words in the public opinion triggering event, determining whether the number of sensitive words exceeds the threshold, and expanding the context loading range when the determination result is yes.

[0009] As a preferred technical solution of the present invention, step four further includes: controlling the system to count the number of sensitive words n in the public opinion triggering event, and setting the context loading range to A1 times the standard value, where A1=n / n0×c, n0 is a pre-input constant, and c is a pre-input correction coefficient.

[0010] As a preferred technical solution of the present invention, step four further includes: after the click is completed, the control system triggers a confirmation event, decrements the number of public opinion hits by one, and synchronizes with the records built into the control system.

[0011] As a preferred technical solution of the present invention, the control system pre-divides the account into a main account and a sub-account, wherein the main account is used to modify the sensitive word grouping and dictionary, the standard value of the context loading range, and the values ​​of n0 and c.

[0012] As a preferred technical solution of the present invention, step one further includes: the control system is used to capture the schedule and remove sub-accounts that are not in a scheduled state from the selection range of the first person to handle the matter.

[0013] An enterprise instant messaging public opinion monitoring and conversation real-time reconstruction device, applicable to the aforementioned enterprise instant messaging public opinion monitoring and conversation real-time reconstruction method, includes: a control module, which is used to carry a control system, and the module is electrically connected to a storage module and an alert module. The storage module is used to record sensitive word groups and a dictionary, and the alert module is used to trigger and issue effective alerts.

[0014] The beneficial effects of this invention are as follows: By randomly selecting the first person to handle the issue and using an idempotent key for verification, the person to handle the issue is identified, while duplicate reminders are avoided. This ensures timely handling and reduces the probability of repeated reminders for the same event. By having the control system count the number of sensitive words in public opinion trigger events, determine whether the number of sensitive words exceeds the threshold, and if the result is yes, expand the context loading range to set the capture range so that when the number of sensitive words is large and the situation is abnormal, the context can be fully invoked, so that the person in charge can fully understand the cause and effect of the conversation and prevent incomplete information retrieval. By using the control system to capture the work schedule and removing sub-accounts that are not on duty from the selection range of the first responder, this addresses the situation in enterprise public opinion monitoring where the responder may not be on duty, resulting in the inability to handle issues in a timely manner. Attached Figure Description

[0015] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0018] Please see Figure 1 A method for monitoring public opinion and reconstructing conversations in real time for enterprise instant messaging includes the following steps: Step 1: The control system uniformly captures messages from chats with customers and group chats, and extracts and organizes the captured messages into text, completing the standardized preprocessing of the captured information. Before executing step one, you need to set up sensitive word groups and dictionaries in the storage module. Sensitive word groups are added, deleted and maintained by the main account. If necessary, sensitive words are deduplicated. Step 2: Control the system to perform sensitive word matching; match the text sequence with the sensitive word group and the dictionary. If a match is found, generate a standardized public opinion trigger event record and increment the public opinion hit count. The record should include at least: member unique identifier, target session identifier, message ID, session ID, sending time, session type, sender, hit keywords and the group to which they belong. Specifically, the chat system either has built-in IDs or is uniformly assigned to all session names and messages by the control system, and assigns identifiers to all personnel accessing the chat system. For example, member A's identifier is 100001, member B's identifier is 100002, member C's identifier is 100003, the ID of the session containing members A and B is 200001, the ID of the session containing members A, B, and C is 200002, the ID of the first message in session 200001 is 20000100001, and so on. In addition, each message is sent with its own timestamp and sender information; After the number of triggered public opinion events is incremented by one, proceed to step three: Step 3: Execute routing strategy for public opinion triggering events: Determine if the scenario is a group chat. If the result is yes, randomly select a member of the enterprise as the first person to handle the situation. At the same time, use idempotent keys to verify that only one valid reminder is triggered. Then, for the first person to handle the situation, create an entry point for the session where the public opinion triggering event is located in the workbench and wait for the first person to handle the situation to click. After the click is completed, proceed to step 4. Specifically, when using idempotency verification, verification can be performed based on one or a combination of the target session identifier, message ID, and hit keywords, with idempotency keys preventing duplicate reminders. Step 4: Reconstruct the session, focusing on the chat message that triggered the sensitive word. Reload the context backtracking. The loading range is confirmed by the control system. The first person to handle the situation can view the complete context containing the sensitive word through the backtracked context. By randomly selecting the first responder and using idempotent keys for verification, the responder is identified, duplicate reminders are avoided, and the timeliness of processing is ensured while reducing the probability of repeated reminders for the same event.

[0019] In some cases, a single message may contain a large number of sensitive words, which indicates that the situation is quite abnormal. If the first person to handle the situation needs to take action, they should make full use of the context to ensure that the person handling the situation fully understands the cause and effect of the conversation. Therefore, step four also includes: controlling the system to count the number of sensitive words in public opinion triggering events, determining whether the number of sensitive words exceeds a threshold, and expanding the context loading scope if the determination result is yes. Specifically, the control system counts the number of sensitive words n in the public opinion triggering events, and sets the context loading range to A1 times the standard value, where A1 = n / n0 × c, n0 is a pre-input constant, c is a pre-input correction coefficient, and A1 is the natural number rounded up from the result of n / n0 × c. When the number of sensitive words n is large and A1 is large, the control system sets the context loading range to A1 times the standard value, thereby expanding the context loading range accordingly. By having the control system count the number of sensitive words in public opinion trigger events, and determine whether the number of sensitive words exceeds a threshold, and if the result is yes, the context loading range is expanded to set the capture range so that when there are many sensitive words and the situation is abnormal, the context can be fully invoked, so that the person in charge can fully understand the cause and effect of the conversation and prevent incomplete information retrieval.

[0020] Since the hit count is incremented by one if a hit is achieved, the record needs to be cleared after the click is completed. The control system triggers a confirmation event, decrements the hit count by one, and synchronizes it with the record built into the control system. Meanwhile, the control system sets a counter for each entry point waiting to be clicked by the first person to handle the situation. When a first person to handle the situation is designated, the counter on the first person's entry point interface is incremented by one. When the first person to handle the situation clicks on the entry point, the counter is decremented by one.

[0021] To facilitate internal management, the control system pre-divides accounts into primary accounts and sub-accounts. The primary account is used to modify sensitive word groups and dictionaries, standard values ​​of context loading range, and the values ​​of n0 and c.

[0022] In some cases, some employees of a company are not at their posts or are not on duty. If these people are assigned as the first responders, it cannot be guaranteed that the public opinion can be handled. Therefore, the control system is used to capture the duty schedule and remove sub-accounts that are not on duty from the selection range of the first responders. The control system has a pre-set selection pool of first disposal personnel. Every fixed time period, in this embodiment, the time period is one hour. Then, the selection pool of first disposal personnel is modified according to the schedule. When a sensitive word is hit, the first disposal personnel is selected from the selection pool of first disposal personnel. By using the control system to capture the work schedule and removing sub-accounts that are not on duty from the selection range of the first responder, this addresses the situation in enterprise public opinion monitoring where the responder may not be on duty, resulting in the inability to handle issues in a timely manner.

[0023] The present invention also provides an enterprise instant messaging public opinion monitoring and conversation real-time reconstruction device, applicable to the above-mentioned enterprise instant messaging public opinion monitoring and conversation real-time reconstruction method, comprising: a control module, the control module being used to carry a control system, the control module being electrically connected to a storage module and an alert module, the control module being used to record sensitive word groups and a dictionary, the storage module being used to record sensitive word groups and a dictionary, and the alert module being used to trigger and issue effective alerts.

[0024] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for monitoring public opinion and reconstructing conversations in real-time for enterprise instant messaging, characterized in that: Includes the following steps: Step 1: The control system uniformly captures messages from chats with customers and group chats, and extracts and organizes the captured messages into text, completing the standardized preprocessing of the captured information. Step 2: Control the system to perform sensitive word matching; match the text sequence with the sensitive word group and the dictionary. If a match is found, generate a standardized public opinion trigger event record and increment the public opinion hit count. The record should include at least: member unique identifier, target session identifier, message ID, session ID, sending time, session type, sender, hit keywords and the group to which they belong. Step 3: Execute routing strategy for public opinion triggering events: Determine if the scenario is a group chat. If the result is yes, randomly select a company member as the first person to handle the situation. At the same time, use idempotent keys to verify and trigger only one valid reminder. Then, for the first person to handle the situation, create an entry point for the session where the public opinion triggering event is located in the workbench and wait for the first person to handle the situation to click. After the click is completed, proceed to step 4. Step 4: Reconstruct the session, focusing on the chat message that triggered the sensitive word, and perform context backloading. The loading scope is confirmed by the control system. Step four also includes: controlling the system to count the number of sensitive words in the public opinion triggering event, determining whether the number of sensitive words exceeds the threshold, and expanding the context loading range when the determination result is yes.

2. The method for monitoring public opinion and reconstructing conversations in real time for enterprise instant messaging according to claim 1, characterized in that: Step four also includes: controlling the system to count the number of sensitive words n in the public opinion triggering event, and setting the context loading range to A1 times the standard value, where A1=n / n0×c, n0 is a pre-input constant, and c is a pre-input correction coefficient.

3. The method for monitoring public opinion and reconstructing conversations in real time for enterprise instant messaging according to claim 1, characterized in that: Step four also includes: after the click is completed, the control system triggers a confirmation event, decrements the number of public opinion hits by one, and synchronizes with the records built into the control system.

4. The method for monitoring public opinion and reconstructing conversations in real time for enterprise instant messaging according to claim 2, characterized in that: Step one also includes: the control system pre-divides the account into a main account and a sub-account, wherein the main account is used to modify the standard values ​​of sensitive word grouping and dictionary, context loading range, n0 and c.

5. The method for monitoring public opinion and reconstructing conversations in real time for enterprise instant messaging according to claim 1, characterized in that: Step one further includes: the control system is used to capture the shift schedule and remove sub-accounts that are not in a shift status from the selection range of the first person to handle the situation.

6. A device for monitoring public opinion and reconstructing conversations in real time for enterprise instant messaging, characterized in that: The method for monitoring public opinion and reconstructing conversations in enterprise instant messaging according to any one of claims 1 to 5 includes: a control module, which is used to carry a control system, and the control module is electrically connected to a storage module and an alert module. The storage module is used to record sensitive word groups and a dictionary, and the alert module is used to trigger and issue effective alerts.

Images