Customer complaint barrier analysis system based on generative artficial intelligence and method thereof

Abstract

A customer complaint barrier analysis system based on generative artificial intelligence and method thereof are provided, the system includes an automatic summary module, a key information extraction module, a knowledge graph construction module, an information database and a customer complaint classification module based on generative AI. The automatic summary module receives customer complaints and generates customer complaint summaries based on generative AI. The key information extraction module uses NLP technology to extract keywords from the customer complaint summaries. The knowledge graph construction module generates a knowledge graph according to dependency relationship between keywords and equipment installation information. The information database is used for a user to collect and store network and device information in the current network. The customer complaint classification module based on generative AI infers obstacle causes according to the keywords, the knowledge graph, and the network and device information, and classifies the customer complaints.

Description

Technical Field

[0001] This invention relates to a customer complaint obstacle analysis system and method, and more particularly to a customer complaint obstacle analysis system and method based on generative artificial intelligence. Prior Technology

[0002] In handling real-time and complex application scenarios, telecommunications network maintenance faces a challenge when dealing with customer complaints about the same problem, which often involve diverse descriptions from different customers. Currently, the process of maintenance personnel relying on manual experience to determine the cause of the problem cannot summarize the key reasons for the complaint in a timely manner. This not only results in longer troubleshooting times but also fails to provide accurate solutions in a timely and effective manner, leading to poor service quality and customer experience. Summary of the Invention

[0003] This invention provides a customer complaint obstacle analysis system and method based on generative artificial intelligence. It can not only classify and summarize the key reasons for customer complaints in real time, effectively shortening the problem investigation time, but also store key information through semantic vectors, making it convenient for subsequent users to search for historical customer complaints and equipment obstacle information semantically.

[0004] This invention discloses a customer complaint obstacle analysis system based on generative artificial intelligence, comprising an automatic summarization module, a key information extraction module, a knowledge graph construction module, an information database, and a customer complaint classification module based on generative AI. The automatic summarization module receives customer complaints and generates a customer complaint summary based on the customer complaint and generative AI. The key information extraction module is electrically or communicatively connected to the automatic summarization module to receive the customer complaint summary and extract keywords from the summary using Natural Language Processing (NLP) technology. The knowledge graph construction module is electrically or communicatively connected to both the automatic summarization module and the key information extraction module to generate a knowledge graph based on the dependency relationship between keywords and device installation information. The information database allows users to pre-collect and store network and device information currently available on the network. The customer complaint classification module based on generative AI is electrically or communicatively connected to the key information extraction module, the knowledge graph construction module, and the information database. It is used to infer the cause of the fault based on keywords, knowledge graphs, and network and device information, and classify customer complaints according to the cause of the fault.

[0005] The present invention discloses a customer complaint obstacle analysis method based on generative artificial intelligence, comprising: receiving a customer complaint; generating a customer complaint summary based on generative AI and the customer complaint; receiving the customer complaint summary; extracting keywords from the customer complaint summary using natural language processing (NLP) technology; generating a knowledge graph based on the dependency relationship between keywords and device installation information; and inferring the cause of the obstacle based on the keywords, the knowledge graph, and pre-collected network and device information in the current network, so as to classify the customer complaint according to the cause of the obstacle.

[0006] Based on the above, this invention provides a customer complaint fault analysis system and method based on generative artificial intelligence. It can accurately understand the diverse descriptions of the same problem by different customers, and immediately classify and summarize the key reasons for customer complaints. This effectively shortens the problem investigation time, greatly improves processing efficiency, reduces human error, and allows maintenance personnel to respond to customer needs more quickly and provide accurate solutions, thereby improving customer experience and service efficiency. Furthermore, it can store key information through semantic vectors, making it convenient for subsequent users to search for historical customer complaints and equipment fault information semantically.

[0007] To make the above features and advantages of the present invention more apparent and understandable, specific embodiments are described below in conjunction with the accompanying drawings for detailed explanation. Simple Explanation of the Diagram

[0008] Figure 1 is an architecture diagram of a customer complaint obstacle analysis system based on generative AI according to an embodiment of the present invention. Figure 2 is a schematic diagram of a customer complaint obstacle analysis system based on generative AI according to an embodiment of the present invention. Figure 3 is a schematic diagram of a knowledge graph according to a first embodiment of the present invention. Figure 4 is a schematic diagram of a knowledge graph according to a second embodiment of the present invention. Figure 5 is a schematic diagram of a knowledge graph according to a third embodiment of the present invention. Figure 6 is a flowchart of a customer complaint obstacle analysis method based on generative AI according to an embodiment of the present invention. Implementation

[0009] Some embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Component symbols used in the following description are considered to be the same or similar components when they appear in different drawings. These embodiments are only a part of the present invention and do not disclose all possible implementations of the invention.

[0010] Figure 1 is an architecture diagram of a customer complaint obstacle analysis system based on generative AI according to an embodiment of the present invention.

[0011] The customer complaint obstacle analysis system 100 based on generative AI proposed in this invention is mainly used in the classification and processing of customer complaints in telecommunications network maintenance. It helps maintenance personnel quickly grasp the core issues of customer complaints, significantly reduces manual processing time, can accurately understand the diverse descriptions of the same problem by different customers, and can instantly classify and summarize the key reasons for customer complaints, effectively shortening the problem investigation time and greatly improving processing efficiency and accuracy.

[0012] Referring to Figure 1, the customer complaint obstacle analysis system 100 based on generative AI may include an automatic summarization module 110, a key information extraction module 120, a knowledge graph construction module 130, a knowledge graph and semantic vector storage module 140, an information database 160, and a customer complaint classification module 150 based on generative AI.

[0013] The automatic summary module 110 can receive customer complaints, generate customer complaint summaries based on generative AI, and the customer complaint summaries may include information about the customer complaint issues that inform the user.

[0014] The key information extraction module 120 is electrically or communicatively connected to the automatic summarization module 110. It can receive customer complaint summaries and extract keywords from the customer complaint summaries using natural language processing (NLP) technology. The keywords may include information related to network signals or information related to the device. This invention is not limited thereto.

[0015] The knowledge graph construction module 130 is electrically or communicatively connected to the automatic summarization module 110 and the key information extraction module 120, respectively, and can generate a knowledge graph based on the dependency relationship between keywords and equipment installation information.

[0016] The information database 160 allows users to pre-collect and store network and device information currently available on the network. In one embodiment, the user pre-collects network signal information and device installation information related to customer complaints in the current area and stores them in the information database 160. This invention is not limited thereto. In another embodiment, the information database 160 allows users to pre-collect and store device installation information.

[0017] The customer complaint classification module 150 based on generative AI is electrically or communicatively connected to the key information extraction module 120, the knowledge graph construction module 130, the knowledge graph and semantic vector storage module 140, and the information database 160. It can infer the cause of the problem based on keywords, knowledge graphs, and network and device information, and classify customer complaints according to the cause of the problem.

[0018] In one embodiment, the generative AI-based customer complaint classification module 150 can determine whether a customer complaint belongs to a branch of the knowledge graph based on keywords and the knowledge graph. If the customer complaint belongs to a branch of the knowledge graph, the generative AI-based customer complaint classification module 150 can infer the cause of the fault based on the branch and network and device information in the current network, and classify the customer complaint according to the cause of the fault. If the customer complaint does not belong to a branch of the knowledge graph, the process ends.

[0019] The knowledge graph and semantic vector storage module 140 are electrically or communicatively connected to the knowledge graph construction module 130 and the customer complaint classification module 150 based on generative AI, respectively. It can convert keywords into semantic vectors and store each semantic vector as a key-value pair in a key-value database. Users can use semantic similarity to query and retrieve historical customer complaint information from the key-value database. This allows maintenance personnel to search for historical customer complaints and equipment fault information (such as historical customer complaint reports, fault classifications, or solutions) semantically. They can quickly find the fault classification or solution of the customer complaint through retrieval, thereby handling the customer complaint report in a timely and effective manner.

[0020] In one embodiment of the present invention, the automatic summarization module 110, the key information extraction module 120, the knowledge graph construction module 130, the knowledge graph and semantic vector storage module 140, and the customer complaint classification module 150 based on generative AI are implemented using a central processing unit (CPU) or other programmable general-purpose or special-purpose microprocessors. In another embodiment of the present invention, the automatic summarization module 110, the key information extraction module 120, the knowledge graph construction module 130, the knowledge graph and semantic vector storage module 140, and the customer complaint classification module 150 based on generative AI can also be implemented using the same processor and loaded with different modules; the present invention is not limited thereto.

[0021] The automatic summarization module 110, the key information extraction module 120, the knowledge graph construction module 130, the knowledge graph and semantic vector storage module 140, and the customer complaint classification module 150 based on generative AI can be implemented through one of software, firmware, hardware circuits, or any combination thereof. This disclosure does not limit the implementation method of the automatic summarization module 110, the key information extraction module 120, the knowledge graph construction module 130, the knowledge graph and semantic vector storage module 140, and the customer complaint classification module 150 based on generative AI.

[0022] The following text will use the various devices, components, and modules shown in Figure 1 to illustrate how the customer complaint analysis system based on generative AI, as described in the embodiments of the present invention, classifies and processes customer complaints after receiving them.

[0023] Figure 2 is a schematic diagram of a customer complaint obstacle analysis system based on generative AI according to an embodiment of the present invention.

[0024] Please refer to Figure 2. The automatic summary module 110 receives the following customer complaint. In one embodiment, the content of this customer complaint may be, for example: the signal is often poor, indicating that the location is blocked by high-rise buildings, and the coverage in some areas is poor, affecting indoor reception. It is necessary to build another base station to strengthen the signal, but the improvement is limited. Later, the user requested to report the fee dispute, but was unwilling to provide it for fear of leaking personal information. The user stated that he reported the issue through the service center and did not understand why it became an online message. He explained that he was not clear about the store's operating procedures.

[0025] The automatic summary module 110 generates a customer complaint summary based on generative AI and the customer complaint itself. The generative AI can be pre-trained to generate a generative model (e.g., a Transformer model), allowing the customer complaint to generate a summary that informs the user of the complaint's issues or core problems. In this embodiment, the customer complaint summary generated based on generative AI might include, for example: poor signal, building obstruction, poor coverage, limited improvement, or fee dispute.

[0026] The key information extraction module 120 receives the customer complaint summary and uses NLP technology to extract keywords from the customer complaint summary. In this embodiment, keywords may be, for example, "poor signal" or "poor coverage".

[0027] The knowledge graph construction module 130 can generate a knowledge graph related to customer complaint obstacles based on the dependency relationship between the above keywords and the equipment installation information (such as base station equipment installation information within the scope of customer complaints) pre-stored in the information database 160.

[0028] The customer complaint classification module 150 based on generative AI infers the cause of the problem based on the keywords, knowledge graph, and network and device information (such as pre-collecting information related to the user's download speed and whether the base station device has alarm messages), and classifies the customer complaint based on the cause of the problem.

[0029] The following describes, in conjunction with the first to third embodiments, how to classify customer complaints based on keywords, knowledge graphs, and network and device information.

[0030] Figure 3 is a schematic diagram of a knowledge graph according to a first embodiment of the present invention.

[0031] Referring to Figure 3, in the first embodiment, the customer complaint obstacle analysis system 100 uses the knowledge graph 301 shown in Figure 3 to perform root cause analysis on mobile network customer complaint obstacle tickets (i.e., customer complaint reports).

[0032] The customer complaint obstacle analysis system 100 determines whether a customer complaint belongs to a branch or category in the knowledge graph. When the customer complaint belongs to the category of "poor coverage", the customer complaint obstacle analysis system 100 infers the cause of the obstacle based on the branch or category and the network and device information in the current network, and performs a final classification of the customer complaint based on the cause of the obstacle. In this embodiment, the customer complaint obstacle analysis system 100 obtains and determines the user's download speed from the information database 160, and combines it with the knowledge graph 301 to determine whether it belongs to the subcategory "poor download speed". After determining that it belongs to the subcategory "poor download speed", the customer complaint obstacle analysis system 100 will then query the information database 160 to see if there is a 5G base station nearby. If there is no 5G base station nearby, the customer complaint obstacle analysis system 100 can determine that the customer complaint belongs to the final category of "no 5G service".

[0033] Figure 4 is a schematic diagram of a knowledge graph according to a second embodiment of the present invention.

[0034] Referring to Figure 4, in the second embodiment, the customer complaint obstacle analysis system 100 uses the knowledge graph 401 shown in Figure 4 to perform root cause analysis on mobile network customer complaint obstacle tickets (i.e., customer complaint reports).

[0035] The customer complaint fault analysis system 100 determines whether a customer complaint belongs to a branch or category in the knowledge graph. When a customer complaint belongs to the customer complaint categories "long-term poor reception" and "poor download speed" in the knowledge graph 401, the customer complaint fault analysis system 100 infers the cause of the fault based on the branch or category and the network and equipment information in the current network, and performs a final classification of the customer complaint based on the cause of the fault. In this embodiment, the customer complaint fault analysis system 100 will further search the information database 160 to see if the base station equipment has alarm messages, and then determine whether the customer complaint belongs to the subcategory of "base station failure". Then, through the detailed description of the alarm message, it determines which type of base station equipment failure the customer complaint belongs to, such as a radio frequency module (Remote Radio Unit, RRU) failure, a baseband unit (BBU) failure, or an alarm supervisor unit (ASU) failure.

[0036] Figure 5 is a schematic diagram of a knowledge graph according to a third embodiment of the present invention.

[0037] Referring to Figure 5, in the third embodiment, the customer complaint obstacle analysis system 100 uses the knowledge graph 501 shown in Figure 5 to perform root cause analysis on mobile network customer complaint obstacle tickets (i.e., customer complaint reports).

[0038] The customer complaint fault analysis system 100 determines whether a customer complaint belongs to a branch or category in the knowledge graph. If the customer complaint belongs to the categories "poor coverage" and "poor download speed" in the knowledge graph 501, the customer complaint fault analysis system 100 infers the cause of the fault based on the branch or category and the network and device information in the current network, and performs a final classification of the customer complaint based on the cause of the fault. In this embodiment, the customer complaint fault analysis system 100 obtains and further judges the timing from the information database 160. If this situation has occurred for six consecutive months, it is determined to belong to the subcategory "long-term poor reception". Then, the customer complaint fault analysis system 100 will query the customer complaint location from the information database 160 to see if the customer complaint can be handled by "adjusting the antenna configuration" and "adding a transponder".

[0039] The following explanation, in conjunction with Figure 6, illustrates how the Generative AI-based Customer Complaint Obstacle Analysis System 100 categorizes customer complaints based on keywords, knowledge graphs, and network and device information.

[0040] Figure 6 is a flowchart of a customer complaint obstacle analysis method based on generative AI according to an embodiment of the present invention.

[0041] Please refer to Figure 6. In step S601, the automatic summary module 110 receives customer complaints and generates customer complaint summaries based on generative AI and the customer complaints.

[0042] In step S602, the key information extraction module 120 extracts keywords from the customer complaint summary.

[0043] In step S603, the knowledge graph construction module 130 generates a knowledge graph based on the dependency relationship between keywords and equipment installation information.

[0044] In step S604, the customer complaint classification module 150 based on generative AI determines whether the customer complaint belongs to one of the branches in the knowledge graph.

[0045] If it is determined that the customer complaint belongs to one of the branches in the knowledge graph, in step S605, the customer complaint classification module 150 based on generative AI infers the cause of the fault based on the branch and the network and device information in the current network, and classifies the customer complaint according to the cause of the fault.

[0046] If it is determined that the customer complaint does not belong to any branch in the knowledge graph, the process ends in step S606.

[0047] Based on the above, this invention provides a customer complaint fault analysis system and method based on generative artificial intelligence. It can accurately understand the diverse descriptions of the same problem by different customers, and immediately classify and summarize the key reasons for customer complaints. This effectively shortens the problem investigation time, greatly improves processing efficiency, reduces human error, and allows maintenance personnel to respond to customer needs more quickly and provide accurate solutions, thereby improving customer experience and service efficiency. Furthermore, it can store key information through semantic vectors, making it convenient for subsequent users to search for historical customer complaints and equipment fault information semantically.

[0048] Although the present invention has been disclosed above by way of embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make some modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0049] 100: Customer Complaint Barrier Analysis System Based on Generative AI 110: Automatic Summary Module 120: Key Information Extraction Module 130: Knowledge Graph Construction Module 140: Knowledge Graph and Semantic Vector Storage Module 150: Customer Complaint Classification Module Based on Generative AI 160: Information Database 301, 401, 501: Knowledge Graph S601, S602, S603, S604, S605, S606: Steps

Claims

1. A customer complaint obstacle analysis system based on generative artificial intelligence, comprising: An automatic summary module is used to receive customer complaints and generate a customer complaint summary based on the customer complaint using generative AI. A key information extraction module, electrically or communicatively connected to the automatic summarization module, receives the customer complaint summary and extracts keywords from it using Natural Language Processing (NLP) technology. A knowledge graph construction module, electrically or communicatively connected to both the automatic summarization module and the key information extraction module, generates a knowledge graph based on the dependency relationship between the keywords and device installation information. An information database is provided for users to pre-collect and store network and device information currently available on the network. A customer complaint classification module based on generative AI is electrically or communicatively connected to the key information extraction module, the knowledge graph construction module, and the information database, respectively, and infers the cause of the problem based on the keywords, the knowledge graph, and the network and device information, classifying the customer complaint based on the cause of the problem.

2. The customer complaint obstacle analysis system as described in claim 1, wherein the customer complaint obstacle analysis system further comprises: The knowledge graph and semantic vector storage module are electrically or communicatively connected to the knowledge graph construction module and the customer complaint classification module based on generative AI, respectively, to convert the keyword into semantic vectors and store each semantic vector as a key-value pair in a key-value database, so that the user can query and obtain historical customer complaint information from the key-value database using semantic similarity.

3. The customer complaint fault analysis system as described in claim 1, wherein the keyword includes information related to network signals or information related to equipment.

4. The customer complaint barrier analysis system as described in claim 1, wherein the customer complaint summary includes the customer complaint issue that informs the user of the customer complaint.

5. The customer complaint obstacle analysis system as described in claim 1, wherein the operation of the customer complaint classification module based on generative AI to infer the cause of the obstacle based on the keyword, the knowledge graph, and the network and device information, and to classify the customer complaint based on the cause of the obstacle, further includes: The generative AI-based customer complaint classification module is further used to determine whether the customer complaint belongs to a branch of the knowledge graph based on the keyword and the knowledge graph. In response to the customer complaint belonging to that branch of the knowledge graph, the generative AI-based customer complaint classification module is further used to infer the cause of the fault based on the branch and the network and device information in the current network, so as to classify the customer complaint based on the cause of the fault.

6. A method for analyzing customer complaint barriers based on generative artificial intelligence, comprising: Receive customer complaints and generate a customer complaint summary based on generative AI. The system receives the complaint summary and extracts keywords from it using Natural Language Processing (NLP) technology. It then generates a knowledge graph based on the dependency relationship between the keywords and device installation information. Finally, it infers the cause of the fault based on the keywords, the knowledge graph, and pre-collected network and device information in the current network, and classifies the complaint according to the cause of the fault.

7. The customer complaint obstacle analysis method as described in claim 6, wherein the method further includes: The keyword is converted into semantic vectors and each semantic vector is stored as a key-value pair in a key-value database so that the user can use semantic similarity to query and retrieve historical customer complaint information from the key-value database.

8. The customer complaint fault analysis method as described in claim 6, wherein the keyword includes information related to network signals or information related to the device.

9. The customer complaint barrier analysis method as described in claim 6, wherein the customer complaint summary includes the customer complaint issue that informs the user of the customer complaint.

10. The customer complaint obstacle analysis method as described in claim 6, wherein the step of inferring the cause of the obstacle based on the keyword, the knowledge graph, and pre-collected network and device information in the current network, and classifying the customer complaint based on the cause of the obstacle, further includes: Based on the keyword and the knowledge graph, determine whether the customer complaint belongs to a branch of the knowledge graph; And in response to the fact that the customer complaint belongs to that branch in the knowledge graph, infer the cause of the fault based on that branch and the network and device information in the current network, and classify the customer complaint according to the cause of the fault.

Images