The Changing Tendency and Association Analysis of Intelligent Coal Mines in China: A Policy Text Mining Study

Abstract

The intellectualization of coal mines provides core technical support for the high-quality development of the coal industry. Intelligent texts, especially intelligent policy documents, play an extremely important role in analyzing the trend of intelligent policies in coal mines. This paper collects more than 50 central and local intelligent coal mine policy texts from recent years. The method of text analysis is a tool used for text mining, and semantic networks are generated; it reflects that the policy mainly focuses on promoting large-scale equipment and platform integration, to promote the overall goal of safe, efficient, and intelligent development of coal mining. By analyzing the high-frequency words of the policy from 2016 to 2022, it reflects that the policy trend mainly goes through the following three stages: firstly, eliminate backward enterprises, encourage coal mine automation and mechanization; then, standardize the basic concept of coal mine intellectualization, carry out the transformation of coal mine intellectualization; and the third stage is to promote the application of key technologies of intellectualization, build intelligent demonstration coal mines and reach the acceptance stage, and promote the further development trend of coal mine intellectualization.

1. Introduction

Promoting the intelligent construction of coal mines is an important method to promote the green, safe, and efficient mining of coal [1,2,3,4]. The coal mine intelligent policy is mainly through the further integration of traditional mining and new generation information technologies such as artificial intelligence and big data. In order to analyze the key points and change trends of the coal mine intelligent policy, it is necessary to screen and collect the text of the coal mine intelligent policy [5,6,7]. In order to speed up the intelligent construction of coal mines, the central and local energy bureaus have issued relevant policies and measures to promote the intelligent development of coal mines. Through the screening of relevant intelligent policies, more than 50 coal mine intelligent construction policies were issued by 11 provinces (cities and autonomous regions) including Shanxi, Guizhou, Henan, Shandong, Inner Mongolia, Shaanxi, and Xinjiang, and finally collected. The collected policy texts mainly point out the guiding ideology, basic principles, main objectives, and main tasks of intelligent development, and the recorded form of all policy contents are relatively unified, which meets the requirements of text analysis. However, the intelligent construction of coal mines is a complex project with multi-systems, multi-fields, and multi-levels [8,9]. The content covered by the policy has the problems of diverse and heterogeneous information and inconsistent description and expression. Therefore, through text feature analysis, co-occurrence analysis, and change trend analysis of the intelligent policy, it is more conducive to systematically reveal its potential laws.

In the context of big data, it is necessary to fully understand the relevant policy requirements of intelligent mines and the development trend of the coal industry in the direction of intelligence [10,11,12,13]. Therefore, it is of great significance to analyze the intelligent policy through text analysis. Some scholars have built an automatic classification model of coal mine accident cases and applied the Word2vec-cgSVM automatic classification model to online coal mine accident case classification [14,15,16,17,18]. Some scholars have studied the theory of hidden dangers of safety production accidents and studied the weak links in the investigation of hidden dangers of coal mine accidents by using the topic model and association rules [19,20,21,22]. Some researchers used the association rule mining model to analyze mine hidden danger data, explored the internal relationship between different hidden dangers, and carried out visual displays [23,24,25,26,27]. Some analyzed the keywords of the policy texts of the coal industry from 2013 to 2018 and their network relations from the four dimensions of capacity adjustment, innovation and efficiency, safe production, and green development [28,29,30]. Some scholars have identified the risk factors of coal safety production through text preprocessing of coal mine accident reports and the combination of text mining, association rule mining, and Bayesian network methods [31].

At present, few scholars use text mining and co-occurrence analysis technology to study the intelligent policy of coal mines. Based on the policy text preprocessing, this paper makes a preliminary analysis of the policy information and analyzes the key objectives and tasks in the policy through the TF-IDF algorithm. For the correlation analysis of the key contents of the policy, the main contents are co-occurrence analysis and centrality analysis by using ROSTCM6 software. The co-occurrence network diagram is optimized by Gephi software, to analyze the correlation contents centered on different factors and realize the visualization of the co-occurrence network. Aiming at the change in policy over time, this paper analyzes the change of high-frequency words in policy text in different years, constructs a relational network diagram, and realizes the visual analysis of policy content over time.

2. Text Feature Analysis of Coal Mine Intelligent Policy Based on Text Preprocessing

2.1. Text Source and Preprocessing

Coal mine intelligence is the core of the high-quality development of China’s coal industry. The intelligent construction of coal mines is inseparable from the guidance of policy top-level design. To analyze the intelligent policy texts, more than 50 intelligent construction policies for coal mines issued by the central government and 11 provinces (cities, autonomous regions) including Shanxi, Guizhou, Henan, Shandong, Inner Mongolia, Shaanxi, and Xinjiang were selected from the websites of the state and local energy bureaus. According to the policy theme, useful content is filtered out and integrated into text documents.

For the selected text content, due to a large amount of data and the diversity of text structure, to ensure the effect of text mining, first of all, the content of unstructured and unmarked text was cleaned, and the words with similar meanings were processed uniformly and standardized, and then the Jieba package was used for word segmentation. To improve the accuracy of word segmentation, a custom thesaurus of coal mine intelligent policy was constructed before word segmentation, and in addition to the actual content of the policy, the thesaurus also covered professional words such as mining engineering and Internet science, to build an intelligent policy thesaurus for coal mines. Meaningless words such as adverbs and function words were added to the stop word list when building the stop word list. The TF-IDF algorithm and the method of calculating high-frequency words were used to extract the feature values of the preprocessed text. Data analysis was carried out through classification, association analysis, and other methods. Finally, the word cloud diagram and association network diagram were used to visualize text results, as shown in Figure 1.

2.2. Word Cloud Display

The central government and relevant provinces and cities have formulated policies for the intelligent development of coal mines and proposed the supply side structural reform of the coal industry. The policy has promoted the integration of the coal mining industry and intelligent technology and improved the intelligence level of coal mining. A large number of words are generated in the policy text after word segmentation, and a large number of “low-quality” words are also generated as the result of word segmentation. Therefore, it is necessary to extract feature words from the text. At present, the methods used to solve text processing problems mainly include Bag of words, TF algorithm, Neural Networks, and TF-IDF algorithm. In the Bag of words method, all the words appearing in the text are regarded as the same, and cannot reflect the importance of different words, so it has a low accuracy. The TF algorithm only takes word frequency as the data representation of words, which is not accurate enough. Word vector representation based on a neural network is popular in text classification. Neural network algorithms include the CBOW and Skip-gram methods. The quality of the neural network model is affected by the data volume, parameter setting, training times, and other factors, and the numerical value cannot directly express the importance of words. The TF-IDF algorithm developed on the basis of the TF algorithm can more accurately express the importance of words. The main idea of the TF-IDF algorithm is that if a word appears frequently in one article and rarely appears in other articles, it is considered that this word or phrase has good discrimination ability, so this study uses this algorithm for keyword extraction.

TF (term frequency) refers to the number of times a word appears in the file. IDF (inverse document frequency) reflects the importance of a word in the document data set. The smaller the number of documents containing a word, the larger the IDF value.

$$T{F}_{ij}=\frac{n_{i,j}}{{\sum }_k{n}_{k,j}}$$

(1)

$$ID{F}_i=\log \frac{\left|D\right|}{\left|\left\{j:t_i\in d_j\right\}\right|}$$

(2)

$$TF-IDF=TF\ast IDF$$

(3)

where $n_{i,j}$ refers to the frequency of word $t_i$ appearing in document $d_j$; ${\sum }_k{n}_{k,j}$ represents the total number of occurrences of all words in document $d_j$; $\left|D\right|$ represents the total number of documents; $\left|\left\{j:t_i\in d_j\right\}\right|$ represents the number of documents containing the word $t_i$.

According to the analysis of policy keywords, the characteristic nouns are coal mine, intelligence, 5G, technology, equipment, working face, Internet, safety and efficiency, robot, and so on. The verbs are manage, integrate, encourage, innovate, and so on. It can be seen that the collection of texts meets the expected objectives. The policy encourages coal mining enterprises to promote the application of relevant intelligent technologies and equipment in the intelligent development of coal mines. It deeply integrates the new generation of information technology with the traditional mining industry, it also puts forward that the low delay and wide connection characteristics of 5G technology should be used to promote the reform of the information system. The unmanned operation of coal mines is promoted by coal mine robots to promote the development of coal mine enterprises in the direction of safety, efficiency, and intelligence.

Table 1. TF-IDF calculation results of coal mine intelligent policy.

Words	TF-IDF Value	Words	TF-IDF Value	Words	TF-IDF Value
intellectualization	0.2340	robot	0.0285	mechanization	0.0145
coal mine	0.2072	cooperation	0.0284	automation	0.0141
5G	0.0833	encourage	0.0255	policy decision	0.0136
monitor	0.0536	early warning	0.0246	check before acceptance	0.0132
platform	0.0442	safe and efficient	0.0242	rock burst	0.0128
equipment	0.0406	disaster	0.0222	digitization	0.0127
working face	0.0361	upgrade	0.0204	fully mechanized mining	0.0125
internet	0.0310	long-range	0.0202	accurate	0.0123
automatic	0.0289	heading face	0.0166	unmanned	0.0121
enterprise	0.0286	promotion of information technology	0.0147	opencast mining	0.0112

shows the statistics of policy text keywords excluding some words with similar meanings and meaningless words.

To intuitively display the key contents in the intelligent text, the word cloud diagram of comparison is drawn through the stylecloud package in Python. The results are shown in Figure 2. In the keyword cloud map of policy text, the font size of the keywords represents the importance of the words in each document.

3. Co-Occurrence Network Analysis and Centrality Analysis of Key Contents of the Intelligent Policy Text

3.1. Co-Occurrence Network Construction

Based on the keywords extracted from the policy text, the important guidance content in the text is further determined to form a simplified text. The simplified text is imported into ROSTCM6 text mining software to extract high-frequency words and co-occurrence matrix. The basic principle of the co-occurrence matrix is to describe the common occurrence of two-word pairs in the same part. Through the study of the co-occurrence network diagram, the hidden text association knowledge is revealed, and the processing results in ROSTCM6 software are imported into Gephi software; after operating, a more optimized co-occurrence network diagram is obtained, as shown in Figure 3. The node in the figure represents the centrality. The larger the node, the greater the centrality, which means that the content of the node occupies a more important position in the text. The line between words indicates that the two words are in the same record. The thicker the line, the higher the weight of the two words appearing at the same time, and the distribution of words in the network diagram indicates the closeness of the relationship between words. The co-occurrence network carries out text analysis on the main technical contents of the intelligent policy and connects the related words, reflecting the internal relationship of the keywords in the main measures. Figure 3 clearly shows the construction of a coal mine intelligent platform and system and the requirements of intelligent technology and equipment. The Column chart in Figure 4 reflects the network modules divided in Gephi. The proportion of each module is 33.65%, 22.12%, 18.27%, 13.46%, 5.77%, 3.85%, and 2.88%, respectively. In Figure 3, different colors are used to distinguish the contents of different modules. The module colors and contents in Figure 3 correspond to those in Figure 4. The modules with different colors in Figure 4 represent different relationship groups. Modules with different colors represent different relationship groups. Module 1 is the core concept of the intelligent coal mine. Module 2 is intelligent roadway support and transportation. Module 3 is the intelligent technology and application direction. Module 4 is the intelligent platform and management system. Module 5 is the intelligent collaborative work of three machines in working face. Module 6 is green mining. Module 7 is intelligent coal mining and heading face. The size of the module calculation results reflects the importance of the module in the whole semantic network.

The generated co-occurrence network diagram not only shows the content of different high-frequency policies and the strength of their links but also reflects the importance of a policy point in the whole coal mine intelligent policy text. Figure 5 shows the co-occurrence set diagram centered on coal mine and intelligence, technology, and equipment.

Figure 5a is a co-occurrence network diagram when the center is coal mine and intelligence, and

Table 2. The main words and their degree in Figure 5a.

Word	Degree
equipment	36
platform	33
big data	15
unmanned	12
information infrastructure	5
safety control system	4
talent	4
green mine	3
intelligent coal mining working face	2
intelligent heading face	2

shows the main words and their degree in Figure 5a, which can reflect the importance of few people and no one in the intelligent coal mine for safe and efficient production, and is related to the platform, equipment, green development, talent policy, and other important content. It shows that the intelligent coal mine policy has certain guidance for the talent training of universities and enterprises, as well as the direction of technical research. The intelligent policy mainly reflects the safety of coal mining, the reliability of technical facilities, the environmental sustainability of energy, the information sensitivity of technical architecture, the service system of the system platform, and the combination of system intelligence and humanistic wisdom. It can be seen that the co-occurrence network diagram of coal mine intelligence is highly consistent with the newly proposed 6S concept of coal mine intelligence, where 6S refers to safety, security, sustainability, sensitivity, service, and smartness.

Intelligent equipment and technology and intelligent platform architecture occupy a very important position in the policy text. Figure 5b is a co-occurrence network diagram when the center is equipment, and

Table 3. The main words and their degree in Figure 5b.

Word	Degree
robot	23
coordination	23
working face	16
unmanned	12
hydraulic support	10
large	8
automation	8
belt conveyor	7
shearer	4
intelligent control	3

shows the main words and their degree in Figure 5b. It can be seen that the equipment is closely related to words such as tunneling, large-scale, automation, few people and no one, robot, shearer, hydraulic support, coordination, intelligent control, and so on. It reflects that the policy promotes the application of intelligent tunneling equipment, memory cutting of shearer, adaptive support of hydraulic support, intelligent transportation of scraper conveyor, and Intelligent Collaborative Control of other equipment, to achieve the construction goals of intelligent heading face, intelligent fully mechanized mining face and other aspects, and promote the development of coal mining towards large-scale, mechanized, and unmanned. By guiding the construction of intelligent platform architecture, the construction goal of an information infrastructure system can be achieved. Figure 5c is the co-occurrence network diagram when the center is the platform, and

Table 4. The main words and their degree in Figure 5c.

Word	Degree
remote control	19
big data	15
signal communication	11
efficient	7
5G	4
informatization	4
intelligent computing platform	2
cloud platform	2
safety monitoring system	2
precise positioning	1

is the main words and their degree in Figure 5c. It can be seen that the platform is combined with keywords such as remote control, cloud platform, informatization, big data, communication, production system, navigation, image, safety monitoring system, intelligent computing platform, fault diagnosis, and so on. It shows that intelligent coal mines pay attention to the construction of integrated management and control platform, which needs to have the functions of production scheduling, safety supervision, personnel communication, positioning and navigation, intelligent operation, fault diagnosis, and so on.

3.2. Centrality Analysis

By importing the semantic network processed in ROSTCM6 into the network analysis software Gephi, the degree centrality, closeness centrality, and betweenness centrality in the intelligent policy semantic analysis network are derived through Gephi. They are listed from high to low according to the centrality and the output results are shown in

Table 5. Analysis results of semantic network centrality of coal mine intelligent policy.

Number	Words	Degree	Closeness Centrality	Betweenness Centrality
1	coal mine	51	0.632911	1339.74047
2	intellectualization	42	0.595238	710.14753
3	equipment	36	0.581395	532.344042
4	platform	33	0.568182	561.050618
5	research and development	18	0.492611	82.498955
6	robot	23	0.512821	198.836942
7	big data	15	0.485437	130.907198
8	monitor	18	0.520833	520.785928
9	transport	18	0.502513	314.862122
10	unmanned	12	0.456621	18.648321
$⋮$	$⋮$	$⋮$	$⋮$	$⋮$
111	innovate	13	0.46729	27.234098
112	hydraulic support	10	0.4329	41.257284
113	heading face	8	0.423729	6.373761
114	automation	8	0.446429	8.184273
115	belt conveyor	7	0.3861	33.569469
116	support	2	0.42735	8.467956
117	internet of things	3	0.367647	2.438542
118	talent	4	0.421941	43.160271
119	safety monitoring system	2	0.369004	1.672941
120	fault diagnosis	2	0.374532	0.479553

.

The degree is the most commonly used indicator in centrality calculation, which indicates the number of edges connected with nodes. The higher the degree is, the more important the node is in the network. Closeness centrality reflects the closeness between a node and other nodes; that is, the reciprocal of the sum of the shortest path distance from a point to other nodes. The closer to other nodes, the greater the closeness centrality of the node. Betweenness centrality refers to the number of shortest paths through a node, and it refers to the transit between this node and other nodes. In Table 5, it can be observed that the three central indicators are generally corresponding. The greater degree represents the ability of the word to reflect the theme in the text, such as coal mine, intelligence, equipment, platform, data, and so on. However, the degree is not enough to fully describe the key content of the text; it needs to combine closeness centrality and betweenness centrality. The smaller betweenness centrality does not mean that the content is unimportant, but that the word is important in specific conditions and locations underground. For example, section, tunneling, support, scraper conveyor, and belt conveyor respectively represent the intelligent requirements for underground roadway tunneling, coal mining, support, and transportation. The closeness centrality analyzes the intelligent requirements of these different aspects according to the length of the path.

4. Text Analysis of Intelligent Policies in Recent Years

4.1. Relationship Network Analysis of High-Frequency Words in Intelligent Policies from 2016 to 2022

The text policy was classified according to year, and the change in policy focus direction in different years was analyzed. The text features of various intelligent policies from 2016 to 2022 were determined through word frequency statistical analysis. By segmenting the text content, removing the stop words, and making word frequency statistics, the words with higher word frequency from 2016 to 2022 were selected after removing the interfering words with little significance. There are many of the same representative words in the policies of these years, corresponding to the co-occurrence network constructed above, and there are also some characteristic words that distinguish each year from other years. To more clearly show the relevance and difference of the characteristic words in the policy text of each year, a relevance matrix of high-frequency words was constructed. The incidence matrix data table was input into Gephi to obtain the association network diagram, as shown in Figure 6. Nodes of different colors in Fig. 6 are used to distinguish high-frequency words in different years.

The association network diagram shown in Figure 6 reflects the priorities and common points of planning in different years. In the middle part of the network diagram are high-frequency words commonly appearing in the policy texts from 2016 to 2022. These high-frequency words can reflect the contents that have been attached importance to in the policy and some main contents, such as artificial intelligence, big data, remote control, security monitoring system, and so on; the outer words are specific measures to promote intelligence, such as intelligent transportation, intelligent ventilation, visualization, intelligent auxiliary transportation system, accurate positioning of equipment, and so on.

According to the relationship network diagram in Figure 6, and the change trend of the coal mine intelligent policy in

Table 6. Change trend of coal mine intelligent policy.

Year	Typical Keywords	Trend
2016	automation, eliminate backward enterprises, informationize, equipment, modernization, machines replace workers, automatic personnel reduction	Eliminate backward enterprises, put forward the production concept of green development, safety and efficiency, and encourage mechanized replacement and automatic personnel reduction.
2017	mechanization, safety control system, disaster prevention, gas, automatic personnel reduction, machine replacement worker, eliminate
2018	equipment, mechanization, green development, safe and efficient, intellectualization, disaster prevention, modernization
2019	intellectualization, equipment, network security, artificial intelligence, internet of things, internet, tunnel boring machine, hydraulic support, shearer, coordination	Carry out the intelligent transformation of mining machinery and standardize the basic concepts and requirements of intelligent mines.
2020	intellectualization, intelligent demonstration coal mine, platform, subsystem, internet, coal mine robot, intelligent perception, intelligent monitoring
2021	intellectualization, equipment, intelligent demonstration coal mine, coal mine robot, intelligent monitoring and early warning, precise positioning, virtual reality, geological big data cloud platform	Promote the application of key technologies of coal mine intellectualization, build intelligent demonstration coal mines, and the intellectualization of coal mines has further developed.
2022	intellectualization, intelligent demonstration coal mine, upgrade, information infrastructure, check before acceptance, intelligent ventilation, integrated support system

, from 2016 to 2022, the requirements of the policy for coal mining enterprises have gradually developed from automation and mechanization to intelligence. From 2016 to 2018, the main content of the policy was to eliminate backward enterprises and provide the basis for the promotion of later intelligent construction. The policy mentions less about intelligent content, mainly encourages the development of coal mine mechanization and automation, and puts forward the concept of green development, safety, and efficiency. The policies of 2019 and 2020 have requirements for the construction of relevant technical equipment, Internet, and big data platform, but the more high-frequency words are still shearer, road header, hydraulic support, artificial intelligence, informatization, and so on. The main content is the intelligent transformation of mine machinery and the basic conceptual requirements of mine intelligence. By 2021 and 2022, a large number of precise, systematic, and intelligent words appeared in the policy. Such as precise positioning of the equipment, high-precision geological model, precise geological exploration, high efficiency, virtual reality, localization, comprehensive support system, information infrastructure, and other words. This is largely due to the basic application of various intelligent technologies in coal mine equipment and information network, which has improved the various functional systems of the mine, it also promotes the development of the intelligent mine to a safer and more efficient direction with fewer people or no one. In addition, the high-frequency words in 2021 and 2022 also include acceptance, scoring, and other words. It shows that the application of intelligent technology has become more and more mature in many mines, forming an intelligent scoring standard system for coal mines, and reaching the acceptance stage.

4.2. Application of Intelligent Key Technologies in Mines

To analyze the response of local coal mines to the intelligent policy, it is necessary to analyze the application cases of corresponding technologies in coal mines and enterprises. Huawei and Luomo group applied 5G technology to unmanned mining equipment in the Sandaozhuang mining area. Beijing Tage IDriver Technology together with Baotou Steel Group jointly built an unmanned intelligent mine project under the condition of a 5G network. So far, there has not been a full-featured commercial deployment case of mining 5G network architecture. It is an urgent problem to combine the 5G communication system with the special application scenarios of the mine. With regard to automatic driving technology, the Beijing easy control intelligent driving company will invest in 12 driverless vehicles in Kuangu District in 2020. In addition, NORINCO, Komatsu, and other enterprises have very mature applications in unmanned driving in mining areas. However, at present, there are some problems, such as a small wireless coverage radius and high cost, which still need a short-time delay and a highly reliable wireless communication system for mining. As for the coal mine robot, CITIC Heavy Industry applied an inspection robot underground to ensure the stable operation of a belt transportation system. The independent research and development of a coal mine enterprise in Anhui Province can replace the 12 positions in the catalog of key research and development of coal mine robots. In addition, there are corresponding application cases of big data, industrial Internet, virtual reality, and other technologies in coal mines. It can be seen that coal mining enterprises all over the country have a very positive response to the coal mine intelligence policy.

5. Conclusions

This paper collects more than 50 central and local policies on coal mine intellectualization. Text mining technology is used to analyze intelligent policy text, explore the key policy content of policy encouragement, analyze the co-occurrence relationship of key content, and the change trend of policy text over time. The main results and conclusions are as follows.

(1) The TF-IDF algorithm is used to filter the feature words, and the results are visualized. It reflects that the main characteristic words include intelligence, 5G, platform, equipment, robot, safety, efficiency, innovation, and so on. It shows that the policy encourages coal mine robots and various underground equipment to combine with 5G and other new generation information technologies. The main goal of safe and efficient coal mining can be achieved by promoting the development of coal mine production in the direction of unmanned.

(2) Through co-occurrence analysis and centrality analysis of the key content text, it can be seen that the co-occurrence network and platform centered on coal mine and intelligence, equipment and green development, and talent policy are all connected. It shows that the policy has a certain guiding effect on talent training and research direction. The content of the policy theme is highly consistent with the newly proposed concepts of Safety, Security, Sustainability, Sensitivity, Service, and Smartness. For the co-occurrence network centered on equipment and platform, the policy encourages the development of large-scale and impersonal equipment and the development of an integrated platform. The combination of equipment and platform promotes the development of coal mining towards safe and efficient intelligent mining.

(3) By analyzing the high-frequency words of the coal mine intelligent policy from 2016 to 2022, The key points of the planning in different years are analyzed. It reflects that from 2016 to 2018, the policy focus is to eliminate backward enterprises and encourage coal mines to develop towards automation and mechanization. From 2019 to 2020, the policy focuses on the application and promotion of intelligent-related equipment and the basic requirements of intellectualization. From 2021 to 2022, the policy focuses on the application of the intelligent integrated system. The acceptance and scoring of intelligent coal mines also show that intelligence has been applied to a certain extent in coal mines, and through the intelligent application technology case of the coal mine, it reflects that the coal mine has a very positive response to the intelligent policy.

(4) The focus of this paper is to analyze the key content and co-occurrence relationship of coal mine intelligent policy, and the change trend of intelligent policy content. The semantic tendency of the analysis has a certain guiding effect on the talent training of universities and enterprises. According to the research on the change trend, it is analyzed that the research focus of coal mine intellectualization changes with time, and the results can also be applied to promote the intelligent development of coal mine enterprises. However, research on coal mine intellectualization using a co-occurrence network and trend analysis is only a preliminary discussion. Due to the limitations of similar structure types of policy texts, and because the research mainly stays at the word level rather than the document level mining, it is difficult to obtain good results after using a clustering algorithm. In the future, more types of texts need to be obtained for further discussion. If a more extensive corpus is obtained by means of a web crawler, some different clusters may be obtained after using clustering algorithms with higher accuracy, such as intelligence, accident analysis, and technology development, and more detailed information applied to the intelligent development of coal mines.