Coal mine safety knowledge interaction method and system

Abstract

The invention discloses a coal mine safety knowledge interaction method and interaction system. The interaction method comprises the following steps: S1, constructing a coal mine safety knowledge database; s2, performing text vectorization processing on the user query request to convert the user query request into vector data Q, performing search matching in a vector knowledge database according to the vector data Q, and selecting vector data A with the highest matching degree; s3, generating a corresponding question X according to the selected vector data A with the highest matching degree in combination with a cue word model; the question X is sent to the large language model LLM, and the large language model LLM outputs corresponding safety knowledge according to the question X. According to the method, the problems of high manual dependency, low user question understanding rate and incomplete knowledge composition of a traditional coal mine safety knowledge base can be solved, the construction efficiency and question and answer accuracy of the coal mine safety knowledge base can be improved through an intelligent means, and reliable assistance is provided for coal mine safety production.

Description

Technical Field

[0001] The present invention relates to the technical field of human-computer interaction, and in particular to a coal mine safety knowledge interaction method and interaction system. Background Art

[0002] In recent years, with the rapid development of artificial intelligence technology, the Large Language Model (LLM) has demonstrated powerful capabilities in natural language processing, knowledge extraction, text generation, etc. The Large Language Model (LLM) has learned to understand and generate complex patterns of human language through pre-training on large-scale data sets, and can automatically process and analyze large amounts of text data.

[0003] The coal mine safety knowledge database uses data mining, text mining and other methods to discover the implicit laws, rules and patterns in massive safety data and make them transparent, explicit and visual. It is an effective way to solve the problems of insufficient information processing and processing capabilities and low information cognition efficiency in the coal mine safety management process. Since coal mine safety training knowledge involves a wide range of knowledge, some staff members are unable to effectively remember the relevant knowledge for a long time, resulting in some work being unable to be carried out or implementing incorrect operations. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a coal mine safety knowledge interaction method and interactive system, which can enable staff to quickly understand relevant coal mine safety knowledge, and can help coal mine enterprises and coal mine sites improve the level of safety knowledge management, improve work efficiency, and achieve business quality and efficiency improvement.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a coal mine safety knowledge interaction method, comprising the following steps:

[0006] S1. Build a coal mine safety knowledge database;

[0007] S1.1. Collect a variety of coal mine safety knowledge documents, and segment the coal mine safety knowledge documents to form a plurality of safety knowledge segments;

[0008] S1.2, converting the security knowledge fragment into vector data A through text vectorization, and storing it in the vector knowledge database;

[0009] S2, converting the user query request into vector data Q by text vectorization, searching and matching the vector data Q in the vector knowledge database, and selecting the vector data A with the highest matching degree;

[0010] S3. Generate a corresponding question X based on the selected vector data A with the highest matching degree and the prompt word model; send the question X to the large language model LLM, and the large language model LLM outputs corresponding security knowledge based on the question X.

[0011] In one embodiment, the coal mine safety knowledge document includes: coal mine safety regulations document, coal mine safety technology document, expert experience document, and historical accident notification case document.

[0012] In one embodiment, each of the security knowledge fragments includes: a segment title, segment content, and issues involved in the segment.

[0013] In one embodiment, the text vectorization process adopts the Word2vec model.

[0014] In one embodiment, the vector knowledge database is updated regularly.

[0015] In one embodiment, the query request is input through natural language, and the natural language query request is converted into vector data Q; the vector data Q is matched with the vector data A in the vector knowledge database for similarity, and the vector data A with the highest similarity is selected as the vector data with the highest matching degree with the vector data Q.

[0016] In one embodiment, the selected vector data Q is input into a prompt word model, and the prompt word model generates corresponding prompt words according to the vector data Q and forms a question X in natural language.

[0017] The present invention also provides a coal mine safety knowledge interaction system, which adopts the coal mine safety knowledge interaction method, and the system includes:

[0018] A database construction module is used to collect a variety of coal mine safety knowledge documents, and segment the coal mine safety knowledge documents to form multiple safety knowledge fragments; convert the safety knowledge fragments into vector data A through text vectorization, and store them in a vector knowledge database;

[0019] A data matching module is used to convert the user query request into vector data Q by performing text vectorization processing, and to search and match the vector data Q in the vector knowledge database to select the vector data A with the highest matching degree;

[0020] The data interaction module generates a corresponding question X based on the selected vector data A with the highest matching degree and the prompt word model; sends the question X to the large language model LLM, and the large language model LLM outputs corresponding security knowledge based on the question X.

[0021] The beneficial effect of the present invention is that the coal mine safety knowledge interaction method and interaction system of the present invention divides the content of the coal mine safety knowledge document into multiple safety knowledge fragments by constructing a coal mine safety knowledge database, and converts the safety knowledge fragments into vector data to form a vector knowledge database; and then displays answers to user query requests through matching and a large language model. In this way, the problems of high manual dependence, low user question comprehension rate, and incomplete knowledge structure in the traditional coal mine safety knowledge base can be improved. The present invention can improve the construction efficiency and question-answering accuracy of the coal mine safety knowledge base through intelligent means, and provide reliable assistance for coal mine safety production. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0023] Figure 1 It is a flow chart of the coal mine safety knowledge interaction method of the present invention.

[0024] Figure 2 It is a structural block diagram of the coal mine safety knowledge interaction system of the present invention. DETAILED DESCRIPTION

[0025] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0026] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0027] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0028] like Figure 1 As shown, the coal mine safety knowledge interaction method of the present invention includes the following steps: S1. Construct a coal mine safety knowledge database. S2. Convert the user query request into vector data Q by text vectorization processing, search and match the vector data Q in the vector knowledge database, and select the vector data A with the highest matching degree. S3. Generate the corresponding question X based on the selected vector data A with the highest matching degree combined with the prompt word model; send question X to the large language model LLM, and the large language model LLM outputs the corresponding safety knowledge according to question X. Among them, step S1 includes: S1.1. Collect a variety of coal mine safety knowledge documents, and segment the coal mine safety knowledge documents to form multiple safety knowledge fragments; S1.2. Convert the safety knowledge fragments into vector data A by text vectorization processing, and store them in the vector knowledge database.

[0029] The present invention can improve the problems of high manual dependence, low user question comprehension rate and incomplete knowledge structure in the traditional coal mine safety knowledge base. The present invention can improve the construction efficiency and question-answering accuracy of the coal mine safety knowledge base through intelligent means, and provide reliable assistance for coal mine safety production.

[0030] It should be noted that a large language model (LLM) refers to a deep learning model trained with a large amount of text data, which can generate natural language text or understand the meaning of language text. Large language models can handle a variety of natural language tasks, such as text classification, question and answer, dialogue, etc. Text vectorization refers to representing text information as a vector that can express the semantics of the text, and using a numerical vector to represent the semantics of the text. Text vectorization processing technology includes a one-hot model, a bag of words model, a term frequency-inverse document frequency (TF-IDF), an N-gram model, a word-vector model (Word2vec), and a document-vector model (Doc2vec). For example, in this embodiment, the text vectorization process uses the Word2vec model. The Word2vec model converts uncomputable and unstructured words into computable and structured vectors. The word2vec model assumes that the order of occurrence of words is not concerned. Word2Vec includes two network structures: a continuous bag of words model CBOW (Continues Bag of Words) and a Skip-gram model. After training, the Word2Vec model can establish a mapping relationship between words and vectors, so it can be used to represent the relationship between words. Assuming that the security knowledge fragment contains three sentences, such as "What are the main functions of the monitoring system", "What functions does the monitoring system cover", and "What are the main functions of the monitoring system", three six-dimensional vectors can be obtained by text vectorization of these three sentences: a = (12, 4, 3, 5, 3, 4, 24), b = (12, 4, 2, 1, 6, 5, 20), c = (12, 4, 2, 1, 3, 4, 19). That is, text vectorization is to convert text into multi-dimensional vectors according to certain rules, and the conversion rules can be set according to actual needs.

[0031] In this embodiment, the coal mine safety knowledge documents include: coal mine safety regulations documents, coal mine safety technical documents, expert experience documents, historical accident notification case documents, etc. The collected coal mine safety knowledge documents are deduplicated and duplicate content is deleted. Then, the content of the coal mine safety knowledge documents is segmented to form multiple safety knowledge fragments. Each safety knowledge fragment contains: segment title, segment content, and issues involved in the segment. Then, each safety knowledge fragment is converted into vector data and stored in the vector knowledge database. For example, the vector knowledge database is regularly updated to ensure the timeliness and accuracy of the vector knowledge database.

[0032] In this embodiment, the user's query request is input through natural language, and the query request in natural language is converted into vector data Q; the vector data Q is matched with the vector data A in the vector knowledge database for similarity, and the vector data A with the highest similarity is selected as the vector data with the highest matching degree of the vector data Q. The selected vector data Q is input into the prompt word model, and the prompt word model generates the corresponding prompt word according to the vector data Q and forms a natural language question X. The large language model LLM can display the answer corresponding to question X to the user.

[0033] Assume that the user's query request is "What are the main functions of the monitoring system?" The text is vectorized into vector c. The vector c is compared with the vector data in the vector knowledge database. The cosine similarity between the vector c and each vector in the vector knowledge database is calculated. The cosine similarity value ranges from [-1, 1]. The closer the cosine similarity value is to 1, the more similar the pointing directions between the two vectors are. The vector with the highest cosine similarity value is the vector with the highest matching degree. The cosine similarity value can be used to filter out vector data in the vector knowledge database that is identical / similar to the vector c. The large language model LLM is then used to display the content corresponding to the query request to the user.

[0034] Through this method, workers in coal mines can obtain the safety knowledge they want to know (for example, relevant items of coal mine safety regulations, operating instructions for coal mine equipment, etc.) through inquiry and question-and-answer methods, providing decision-making assistance for underground work.

[0035] like Figure 2 As shown, the present invention also provides a coal mine safety knowledge interaction system, including: a database construction module, which is used to collect a variety of coal mine safety knowledge documents, and segment the coal mine safety knowledge documents to form multiple safety knowledge fragments; the safety knowledge fragments are converted into vector data A by text vectorization processing, and stored in the vector knowledge database. A data matching module is used to convert the user query request into vector data Q by text vectorization processing, and search and match the vector data Q in the vector knowledge database to select the vector data A with the highest matching degree. A data interaction module generates a corresponding question X based on the selected vector data A with the highest matching degree combined with a prompt word model; the question X is sent to the large language model LLM, and the large language model LLM outputs the corresponding safety knowledge according to the question X.

[0036] The data matching module includes a vector conversion unit and a matching unit. The vector conversion unit is used to convert the user query request into vector data Q through text vectorization. The matching unit is used to search and match the vector data Q in the vector knowledge database and select the vector data A with the highest matching degree. The data interaction module includes a question generation unit and a question and answer display unit. The question generation unit is used to generate the corresponding question X based on the selected vector data A with the highest matching degree combined with the prompt word model; the question and answer display unit is used to send question X to the large language model LLM, and the large language model LLM outputs the corresponding security knowledge based on question X.

[0037] In summary, the coal mine safety knowledge interaction method and interaction system of the present invention, by constructing a coal mine safety knowledge database, divides the content of the coal mine safety knowledge document into multiple safety knowledge fragments, and converts the safety knowledge fragments into vector data to form a vector knowledge database; and then displays answers to user query requests through matching and a large language model. In this way, the problems of high manual dependence, low user question comprehension rate, and incomplete knowledge structure existing in the traditional coal mine safety knowledge base can be improved. The present invention can improve the construction efficiency and question-answering accuracy of the coal mine safety knowledge base through intelligent means, and provide reliable assistance for coal mine safety production.

[0038] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A coal mine safety knowledge interaction method, characterized in that: The following steps are involved: S1. Build a coal mine safety knowledge database; S1.

1. Collect a variety of coal mine safety knowledge documents, and segment the coal mine safety knowledge documents to form a plurality of safety knowledge segments; S1.2, converting the security knowledge fragment into vector data A through text vectorization, and storing it in the vector knowledge database; S2, converting the user query request into vector data Q by text vectorization, searching and matching the vector data Q in the vector knowledge database, and selecting the vector data A with the highest matching degree; S3. Generate a corresponding question X based on the selected vector data A with the highest matching degree and the prompt word model; send the question X to the large language model LLM, and the large language model LLM outputs corresponding security knowledge based on the question X.

2. The method for interacting with coal mine safety knowledge according to claim 1, characterized in that: The coal mine safety knowledge documents include: coal mine safety regulations documents, coal mine safety technology documents, expert experience documents, and historical accident notification case documents.

3. The method for interacting with coal mine safety knowledge according to claim 1, characterized in that: Each of the security knowledge fragments includes: segment title, segment content, and issues involved in the segment.

4. The method for interacting with coal mine safety knowledge according to claim 1, characterized in that: The text vectorization process adopts the Word2vec model.

5. The method for interacting with coal mine safety knowledge according to claim 1, characterized in that: The vector knowledge database is updated regularly.

6. The method for interacting with coal mine safety knowledge according to claim 1, characterized in that: The query request is input through natural language, and the natural language query request is converted into vector data Q; the vector data Q is matched with the vector data A in the vector knowledge database for similarity, and the vector data A with the highest similarity is selected as the vector data with the highest matching degree of the vector data Q.

7. The method for interacting with coal mine safety knowledge according to claim 6, characterized in that: The selected vector data Q is input into the prompt word model, and the prompt word model generates corresponding prompt words according to the vector data Q and forms a natural language question X.

8. A coal mine safety knowledge interactive system, characterized in that: The method for interacting with coal mine safety knowledge according to any one of claims 1 to 7 is adopted, wherein the system comprises: A database construction module is used to collect a variety of coal mine safety knowledge documents, and segment the coal mine safety knowledge documents to form multiple safety knowledge fragments; convert the safety knowledge fragments into vector data A through text vectorization, and store them in a vector knowledge database; A data matching module is used to convert the user query request into vector data Q by performing text vectorization processing, and to search and match the vector data Q in the vector knowledge database to select the vector data A with the highest matching degree; The data interaction module generates a corresponding question X based on the selected vector data A with the highest matching degree and the prompt word model; sends the question X to the large language model LLM, and the large language model LLM outputs corresponding security knowledge based on the question X.

Images