Guided question and answer alarm receiving and processing method based on NLP speech recognition and AI agent technology

Abstract

The invention relates to the technical field of public safety emergency processing, and discloses an alarm receiving and processing method based on NLP voice recognition and AI agent guided question answering, a user inquires a reporter by means of a guided question answering function of a system or a telephone robot and records the answer made by the reporter in real time, the system calls NLP voice recognition through an interface, and the response of the reporter is recorded in real time. Identified contents are submitted to an AI agent for extraction, abstracting, summarization and structuralization of main contents, and the main contents are displayed to a user and serve as important basic information elements of a case. After the whole alarm receiving and processing process is finished, related content is automatically extracted, a record is generated and printed, the alarm receiving and processing time of a single case can be shortened from 1-3 days to less than 1 hour, and the case handling period is shortened.

Description

Technical Field

[0001] The present invention relates to the technical field of public safety emergency processing, and specifically to a method for receiving and handling police calls based on NLP speech recognition and guided question-answering of AI intelligent agent technology. Background Art

[0002] Limitations of existing technologies: Although some regions have introduced some information technology to assist in police response, the processing of voice information is mostly limited to simple voice-to-text conversion, lacking in-depth understanding of semantics and intelligent interaction, and the solutions are mostly single-node tool-based rather than systemic solutions. Police response cannot be standardized, and the same case data elements are operated and verified by multiple systems, resulting in inconsistent data across multiple links and systems; interrogation records cannot be quickly prepared, which restricts the advancement of subsequent case handling processes, resulting in tight police resources and long case handling cycles. Summary of the invention

[0003] 1. Technical issues to be resolved

[0004] In view of the shortcomings of the existing technology, the present invention provides a method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer, which improves the efficiency and accuracy of receiving police calls, ensures rapid and accurate police dispatch, and enhances the ability to handle public safety emergencies, with remarkable results:

[0005] 1.NLP speech recognition can solve bottlenecks in police work and shorten the case handling cycle.

[0006] 2. Guided Q&A, intelligent and automatic generation of interrogation records, standardized police handling, and improved data accuracy. The time for handling a single case is shortened from 1-3 days to 1 hour;

[0007] 3. AI intelligent body recognition factors are integrated and the accuracy of inquiry factors is increased to 100%.

[0008] (II) Technical solution

[0009] 1. System architecture: The police receiving and handling platform of the present invention includes a user interface layer, a data collection layer, a basic service layer, a data processing layer, and an application service layer.

[0010] User interface layer: It includes three access methods: 96110 alarm reception, police station alarm reception and on-site alarm response, providing a unified interactive entrance for alarm reception in different scenarios. Users can communicate with the system through this interface, and the system can also feedback the processing results to users.

[0011] Data collection layer: Telephone robots and recording devices are used for data collection, supporting data collection in three scenarios: 96110 call reception, police station call reception, and on-site police dispatch. The telephone robot can automatically conduct preliminary communication with the caller and collect key information; the recording device records the entire call reception and inquiry process to ensure data integrity.

[0012] Basic service layer: Integrate NLP speech recognition and AI agents. NLP speech recognition includes speech-to-text and text-to-speech, and uses deep learning algorithms to achieve efficient and accurate speech and text conversion; AI agents include text understanding, logic understanding, intent understanding, and structured information, and can deeply analyze text content and understand the intention of the caller.

[0013] Data processing layer: responsible for data collection, data association, data extraction, data labeling, data comparison and data integration. Data comparison is to compare certain information such as the crime scene and bank card opening bank after the reporter's answer is identified with the dictionary data in the system to ensure the accuracy of the information.

[0014] Application service layer: provides functions such as case management, user management and transcript generation. Case management can classify, store and query case information; user management is used to manage the information of police officers and case reporters; transcript generation automatically generates interrogation transcripts based on structured information.

[0015] 2.NLP speech recognition module: Using advanced NLP speech recognition technology, the voice of the caller is converted into text in real time, and the text is semantically analyzed. In actual applications, facing the complex speech of the caller with an accent, it is more accurate than the traditional speech-to-text conversion. At the same time, the model fully learns the grammatical rules, semantic connotations and pragmatic habits of natural language. In the process of receiving the alarm, when the caller's expression is not clear, the language model can infer the possible accurate expression based on the recognized content and learned language habits. For example, the caller said "I was robbed of my bag in that square, just now, the person ran over there, oh, I can't explain it clearly", the language model infers that the caller may want to express "I was robbed of my bag in [specific square name] just now, and the person who robbed the bag ran to [specific direction]" based on the recognized content such as "square", "robbed bag", "just now", and the common way of describing events in daily language habits, and provides this inference result to the caller, assisting the caller to accurately understand the alarm content, solve the problem of bottlenecks in police handling, and shorten the case handling cycle.

[0016] 3. AI agent module: After receiving the alarm information, the AI ​​agent automatically generates targeted guiding questions based on the identified key information and automatically structures the relevant information. For example, the case type is preliminarily determined based on the keywords of the event described by the caller. If it contains keywords such as "deception", "routine", "trap", and "false", it is preliminarily determined to be a fraud case, and then the relevant key information is asked according to the preset rules. When encountering complex and ambiguous situations, such as when the event described by the caller involves a mixture of multiple methods of committing crimes, the agent can ask and guide according to the caller's answer. For example, when the caller mentioned being defrauded but did not specify the amount of fraud, the agent further inquires based on the dialogue strategy library and reinforcement learning model to ensure that accurate and complete information is obtained. All data elements are obtained in one inquiry, which solves the problem that the traditional manual police reception and handling method is prone to missing case handling elements and greatly shortens the time to obtain alarm information. At the same time, the AI ​​agent will automatically integrate and structure the information obtained by the NLP speech recognition module to form a complete alarm information report, store it in the database, and generate a transcript in a predetermined format. Avoid asking the same question multiple times and prolonging the case handling cycle. The traditional manual police call-handling process has been optimized, with police officers from the police station notifying the victim to go to a designated location for questioning and making an interrogation record; or responding to a crime scene, where the police officer will inquire about the victim's case-related information and make another appointment to go to the local police station to make an interrogation record and file a case. All case information is directly integrated during the call-handling process, which improves case handling efficiency. 50% of the front-line police force has been AI-powered, saving a large amount of police force in business brigades and police stations, which can be used for police activities that require more manpower, such as investigation, arrest, and security duties.

[0017] 4. Automatically generate transcripts: When the system is working, it first obtains data elements through the NLP voice recognition module, and then hands the recognition content to the AI ​​intelligent body module for structured processing, covering the time of questioning, identity information of the person being questioned, amount of money defrauded, process of the case, method of fraud, etc. The structured information is then integrated into a complete alarm information report and stored in the database, and finally the transcript is automatically generated and printed based on this. For police stations receiving calls and on-site police dispatch, the recorder only needs to turn on the recording equipment throughout the process, and after the call is received and handled, the printed transcript will be handed over to the reporter for signature and seal; for 96110 calls, the police receiver can just check the transcript generated by the platform.

[0018] After the reporter is questioned, the application service layer automatically processes the information and generates transcripts in the following steps:

[0019] S1: Use NLP speech recognition to convert the reporter's voice response into text, and use its advanced technology to ensure accurate and efficient conversion.

[0020] S2: With the help of the text understanding and information extraction capabilities of the AI ​​agent, the key content in the reporter's written answer is automatically extracted to achieve accurate extraction.

[0021] S3: Automatically structure key content according to the preset structural framework, classify and organize key information to make its logical structure clear.

[0022] S4: Template matching and content filling are used to generate a complete interrogation transcript based on structured information and fill in the pre-designed transcript template.

[0023] (II) Advantages of the present invention

[0024] Compared with the prior art, the present invention provides a method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer, which has the following beneficial effects:

[0025] (1) Improve the efficiency of receiving and handling police calls: Through automated voice recognition and intelligent guided question-and-answer, the time to obtain alarm information is greatly shortened, and the time cost of manual questioning and recording is reduced. Regardless of whether the police are on-site, at the police station, or at 96110, the police can prepare a record within 10 minutes, greatly improving work efficiency.

[0026] (2) Improve information accuracy and ensure information consistency: The use of NLP technology and AI agents can more accurately understand the intention of the caller and avoid information errors caused by human misunderstanding or record omissions.

[0027] (3) Standardized police response and improved case handling quality: Complete and accurate police response information is transmitted to police officers in advance, which helps them formulate response strategies in advance, improve police response efficiency and incident handling capabilities, and better protect public safety.

[0028] (4) Resolving bottlenecks in police handling and shortening the case handling cycle: By making transcripts in advance, the time required to secure evidence is shortened from three days to less than one hour.

[0029] (5) Data integration and 50% AI-based frontline police force: This is an innovative practice of the public security organs to "move towards technological police force", which will free up at least 50% of the police force.

[0030] (6) The functions provided by the present invention can improve the quality of interrogation records. By using the system of the present invention, we can better avoid problems such as data omissions, non-standard interrogation processes, and omissions in prescribed actions caused by manual operations. At the same time, if the system achieves the expected results after its construction, its functions can be extended to all aspects of law enforcement and case handling in the global field (such as automatic interrogation by intelligent alarm robots, automatic generation of records, etc.), saving a large number of police forces in business brigades and police stations, and using them for police activities that require more manpower to solve, such as investigation and arrest, security duties, etc., truly realizing digital policing and smart policing. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1A schematic diagram of the structure of a transcript generation method based on NLP speech recognition and AI agent guided question-answering technology of the present invention;

[0032] Figure 2 A flowchart of a transcript generation method based on NLP speech recognition and AI agent guided question-answering technology of the present invention; DETAILED DESCRIPTION

[0033] The following will be combined with the 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 without creative work are within the scope of protection of the present invention.

[0034] like Figure 1 Figure 2 As shown, the present invention proposes a method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer, including a user interface layer, a data collection layer, a basic service layer, a data processing layer and an application service layer.

[0035] During operation, the system uses NLP speech recognition to obtain data elements, and then delivers the recognition content to the AI ​​agent for structured processing, such as the time of questioning, the identity information of the person being questioned, the amount of money defrauded, the process of the case, the method of fraud, etc., and then automatically generates and prints the transcript based on the structured information. For on-site police dispatch or police station call-in situations, the recorder only needs to turn on the recording device throughout the process and hand over the printed transcript to the reporter for signature and seal; for 96110 call-in situations, the police receiver only needs to pay attention to the printed transcript and send it to the police dispatcher.

[0036] The user interface layer includes 96110 alarm reception, police station alarm reception and on-site alarm dispatch.

[0037] The data collection layer covers telephone robots and recording equipment, and supports data collection for three scenarios: telephone alarm reception, police station alarm reception, and on-site police dispatch.

[0038] The basic service layer integrates NLP speech recognition (speech to text, text to speech) and AI agents (text understanding, logic understanding, intent understanding, and structured information).

[0039] The data processing layer is responsible for data collection, data association, data extraction, data labeling, data comparison and data integration. Data comparison is to compare certain information such as the crime scene and bank card opening bank after the reporter's answer is identified with the dictionary data in the system.

[0040] The application service layer provides functions such as case management, user management and transcript generation.

[0041] After the reporter is interviewed, the application service layer automatically processes the information and generates transcripts, including the following steps:

[0042] S1: With the help of NLP speech recognition, the reporter's answer is converted from voice to text, and NLP speech recognition technology is used to ensure the accuracy and efficiency of the conversion;

[0043] S2: Automatically extract key content from the reporter's text answer through the AI ​​agent, and accurately extract key information based on the text understanding and information extraction capabilities of the AI ​​agent;

[0044] S3: Automatically structure key content and integrate key information according to the preset structural framework to give it a clear logical structure;

[0045] S4: Automatically generate transcripts based on structured information. Use template matching and content filling to fill the structured information into a pre-designed transcript template to generate a complete interrogation transcript.

[0046] The overall operation process of the system:

[0047] 1. System initialization: Initialize the NLP speech recognition module and AI agent-guided question-answering module in the police handling platform, and load the trained model and related data.

[0048] 2. Alarm receiving process: When the alarm call comes in, the NLP speech recognition module starts working, converting the speech into text and performing semantic analysis in real time. The AI ​​agent generates guiding questions based on the analysis results and interacts with the caller through voice or text. For example, the caller says "I was defrauded", the NLP module identifies the event type as fraud, and the AI ​​agent immediately asks questions such as "How were you defrauded?" and "How much was the fraudulent amount?", and displays the obtained information in real time on the operator's operation interface, which is also an important basic information element of the case.

[0049] 3. Information processing: After the alarm process is completed, the AI ​​agent will automatically integrate all data elements and generate a transcript in a predetermined format. The operator only needs to send the transcript to the terminal device of the police department or relevant police personnel.

[0050] 4. Police dispatch process: When the police arrive at the scene and handle the incident, they can directly give the transcript to the reporter for signature, avoiding multiple inquiries of the same type of questions, optimizing the police response process, and achieving standardized police response.

[0051] The embodiments of the present invention have been shown and described. Those skilled in the art may make various changes, modifications, substitutions and variations to these embodiments without departing from the principles and spirit of the present invention. The scope of protection of the present invention shall be based on the attached claims and their equivalents.

Claims

1. A method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer, characterized in that: The system architecture includes user interface layer, data collection layer, basic service layer, data processing layer, and application service layer. For the police station's call reception and on-site police dispatch, the recorder only needs to turn on the recording device and give the printed transcript to the reporter for signature and seal. For 96110 call reception, the dispatcher only needs to check the transcript finally generated by the police reception and handling platform.

2. According to claim 1, a method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer, characterized in that: The user interface layer includes 96110 alarm reception, police station alarm reception and on-site alarm dispatch.

3. The method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer according to claim 1, characterized in that: The data collection layer is a telephone robot and recording equipment, which supports data collection in three scenarios: 96110 alarm reception, police station alarm reception, and on-site police dispatch.

4. The method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer according to claim 1, characterized in that: The basic service layer integrates NLP speech recognition and AI agents. NLP speech recognition includes speech-to-text and text-to-speech; AI agents include text understanding, logic understanding, intent understanding, and structured information.

5. The method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer according to claim 1, characterized in that: The data processing layer is responsible for data collection, data association, data extraction, data labeling, data comparison and data integration, wherein data comparison is to compare certain information after the reporter's answer is identified, such as the place of crime, bank account opening bank, etc. with the dictionary data in the system.

6. The method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer according to claim 1, characterized in that: The application service layer provides functions such as case management, user management and transcript generation. Case management can classify, store and query case information; user management is used to manage user information; and transcript generation automatically generates interrogation transcripts based on structured information.

7. The method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer according to claim 1, characterized in that: The NLP speech recognition module adopts a language model based on the Transformer architecture to achieve real-time speech-to-text, text-to-speech and semantic analysis, and accurately identify key information.

8. The method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer according to claim 1, characterized in that: The AI ​​agent adopts a strategy that combines rules with deep learning; the content recognized by the NLP speech recognition module is handed over to the AI ​​agent module for structured processing, including the questioning time, the identity information of the person being questioned, the amount of money defrauded, the process of the incident, the method of fraud, etc., and then the structured information is integrated to form a complete alarm information report, and stored in the database. Finally, a transcript is automatically generated and printed out based on the structured information, realizing targeted guided question and answer and information structured processing and integration, so that the accuracy of the inquiry elements is increased to 100%.

9. The method for receiving and handling police calls based on NLP speech recognition and AI agent-guided question and answer according to claim 1, characterized in that: After the reporter is questioned, the application service layer performs automatic information processing and transcript generation in sequence. The voice is converted into text with the help of NLP speech recognition. The AI ​​agent extracts key content and performs structured processing. Finally, the transcript is generated based on the structured information, including the following steps: S1: With the help of NLP speech recognition, the reporter's answer is converted from speech to text, and the above-mentioned NLP speech recognition technology is used to ensure the accuracy and efficiency of the conversion; S2: Automatically extract key content from the reporter's text answer through the AI ​​agent, and accurately extract key information based on the text understanding and information extraction capabilities of the AI ​​agent; S3: Automatically structure the key content and classify and organize the key information according to the preset structural framework to give it a clear logical structure; S4: Automatically generate transcripts based on structured information. Use template matching and content filling to fill the structured information into a pre-designed transcript template to generate a complete interrogation transcript.

Images