The construction of a multi-platform system for teaching journalism and communication based on big data technology

China’s media industry has experienced the peak of traditional media development and transformation, the vigorous development of new media, the deepening of media convergence, the subversion and innovation of digital technology, and ushered in the era of smart media, but the whole media is also still in a period of great transition of system restructuring, power change and paradigm shift [1-3]. In the face of the great changes in the media environment, the journalism and communication faculties of Chinese universities have also felt the great pressure from the outside, and have made attempts and innovations in talent cultivation to adapt to the media changes in the demand for composite journalism and communication talents [4-6].

Journalism and communication profession is a highly practical and applied discipline, and practice teaching, as an important subsystem of journalism and communication education, plays an increasingly important role in cultivating qualified all-media journalism and communication talents in the era of media convergence [7-9].

In the context of the era of “convergent media”, both traditional newspaper media and radio and television media will gradually be integrated into the “convergent media” system of more powerful digital media. “Convergent media” is a new type of media group that makes full use of media carriers, comprehensively integrates different media such as radio and television, newspapers, etc., seeks common ground while reserving differences, complements each other’s advantages, and builds a new type of media group that is “accommodating in resources, compatible content, integrated in publicity, and inclusive in interests” [10-12]. In the era of “convergent media”, network data has been accumulated, forming a network big data platform [13-15]. Big data is a natural growth from improving production efficiency to a more advanced intelligence stage accumulated by the development of information and communication technology [16-17]. In the era of “convergent media”, the process of news production has changed dramatically, and media organizations have changed the fragmentation caused by the fragmentation of different media and the management of different departments in a complete and integrated process [18-20]. Therefore, how to transform traditional media reporters into all-media reporters is an important issue in the era of “convergent media” [21].

In the teaching of journalism and communication, only by building and making good use of the teaching practice platform, laboratory, and network teaching practice platform of journalism and communication, and carrying out diversified teaching activities, especially the introduction of online self-media teaching platform, can we adapt to the needs of the times and cultivate the all-media senior applied talents required by the journalism and communication major [22-23]. In the professional teaching of journalism and communication, it is an urgent matter to build and make good use of the existing conditions and explore a new mode of journalism and communication teaching in the era of “convergent media” [24-25]. It is necessary to consider the impact of the online “media convergence” platform on the teaching content of journalism and communication, how to enrich the teaching content, what the latest research results in the “convergent media” era have on the teaching methods of the topic, how to improve the teaching content and methods, and how to use the online platform to establish a resource library for journalism and communication teaching [26-28].

Information technology promotes the rapid development of the media industry, and the teaching and training of talents in the corresponding media industry has also received attention, so some scholars have explored the status quo of journalism and communication teaching. Literature [29] uses quantitative and qualitative analysis and research methods to reveal that the teaching of European journalism and communication professional prosthetic courses focuses on theoretical research rather than on professional practice and the application of tools and technologies. Literature [30] elaborated that the current status quo of journalism and communication education in China has been unable to adapt to the needs of the data era, and analyzed the specific path of the integration of big data technology into the teaching of journalism and communication network. Based on the research, it is known that the news communication teaching mode and method are very much to be optimized, so some experts have explored the optimization of news communication teaching around the teaching concept and technology. Based on the concept of competency-based education, [31] builds a collaborative education mechanism from the dimensions of specialties, regions, enterprise goals, faculty, resource sharing, and the construction of teaching platforms, aiming to cultivate journalism and communication application-oriented talents. Based on the practice and questionnaire feedback, the literature [32] confirms that AI teaching-enabled journalism and communication teaching practice is conducive to teaching innovation and has gained the unanimous recognition of teachers and students.

The application of information technology in the field of education has been very common, in which the construction of teaching platform based on big data technology is particularly popular. Literature [33] built an intelligent education platform based on big data technology, which can provide students with career planning services and teaching resources to promote students’ career development and professional ability. Literature [34] designed and evaluated a statistical analysis platform for educational data based on Hadoop, and confirmed that the platform can effectively improve the efficiency of educational data processing and enhance the scientificity of educational decision-making. Literature [35] designed a regional digital teaching resources sharing platform by combining the Internet of Things technology and big data theory, which solved the problems of poor efficiency and damaged integrity of traditional teaching resources uploading and sharing platform to a certain extent. Literature [36] used clustering technology and association rules as the core logic of data analysis to construct a digital and networked teaching platform based on big data technology, with the former favoring personalized teaching resources recommendation and the latter favoring performance analysis and prediction. The research on teaching platforms constructed based on big data technology involves teaching resources sharing, teaching analysis, personalized teaching and students’ career planning, etc. The technology has been relatively mature, while the research on the construction of journalism and communication teaching platforms based on big data technology is blank, so it is necessary to carry out in-depth investigation in this area.

This paper combines DTW-FCM to construct an all-media experimental platform to study the behavioral characteristics behind students’ learning data. It describes the idea of multi-type practical teaching and synthesizes the idea and the fuzzy C-mean clustering algorithm based on DTW to construct an all-media experimental platform with three departments, namely, the collection department, the editorial department and the production department. By analyzing the learning data of students using the all-media experimental platform, students are classified into different types of learners, and further combined with the clustering analysis to derive the behavioral characteristics of different categories of students. The in-depth analysis of students’ behavioral characteristics in each category provides teachers with corresponding ideas for teaching improvement.

2

All-media experimental platform construction

News communication needs to be combined with real-life and work-related content in order to improve the practical teaching level of teachers and help students build a good foundation of skills. This part combines the requirements of multi-type practical teaching and related algorithms to build an all-media experimental platform that provides training platforms and opportunities for journalism learners.

2.1

Multiple types of practical teaching

Multi-type practice teaching is a further innovation based on full-process training. Multi-type can be subdivided into platform type, product type, and content type.

2.1.1

Multi-platform types

The first category, news and information clients, the news and information attributes of these media platforms are prominent, and should focus on the process, skills and thinking methods of media convergence. Which can be subdivided into two categories, one is the new media platform under the traditional news organization; the second is the commercial platform. The second category, social media, WeChat, Jittery, B station, etc., which integrates news information and social interaction and other functions, the teacher should lead the journalism and communication students to change from the role of ordinary users to the role of professional media people. The third category, e-commerce media, teachers can lead students, starting from imitation, from Taobao and other platforms to find top-ranking stores, a comprehensive analysis of the elements of its advertising page, design, marketing techniques, topic settings, etc., in the imitation of the basis, try to independently register and open a store trial operation. The fourth category, tool-based media, can be used by teachers to train students to produce original content on their own, organize user content production, platform product design, and other skills.

2.1.2

Multiple product types

On the one hand, the traditional types of content products in specialized teaching such as text, pictures and videos need to update their concepts and methods with the needs of the new media era, and on the other hand, there is an urgent need to add new emerging types of content product training in practical teaching. Teachers should break the stereotypes, accept and lead students to produce fragmented content, adapt to the presentation of content on different platforms, and strengthen the application of a variety of ways of expression, various types of editing and editing software.

2.1.3

Multiple job types

The supply of single news gathering and editing positions has been greatly reduced, and the types of employment for news and communication graduates are becoming more and more diversified, with a large number of jobs being offered in advertising planning and operation, corporate new media operation, and operation of e-commerce companies, and so on. Practical teaching should no longer be limited to news content, but must also increase the training of advertising, e-commerce, product operation and other positions as much as possible.

2.2

Structure and Functions of the All-Media Experimental Platform

The all-media experimental platform constructed in this paper is divided into three departments, namely, the collection department (see Fig. 1), the editorial department (see Fig. 2) and the production department (see Fig. 3), and each of the three departments has its own function: 1)

The collection department mainly trains students to collect text, pictures, video and audio materials through various channels such as writing, photography, recording, videotaping and website crawling.

2)

The editorial department mainly trains students to use relevant software to process text, pictures, audio, video, multimedia materials, digital resources, thematic production and personalized processing of the collected materials.

3)

The Production Department mainly trains students to use equipment to produce paper books, newspapers and periodicals, electronic books, newspapers and periodicals, radio and TV programs, as well as network, cell phone and outdoor display products. Students’ simulation works can be burned into a CD-ROM, the College to save a copy for future reference, students can copy their own works in order to show their ability to find jobs in the future.

The three departments are interconnected and interdependent, integrating into an all-media experimental platform integrating collection, editing and production. Through the practical training on this platform, students can basically master the all-media communication skills needed to work in today’s press and publishing organizations.

2.3

Design of DTW-based fuzzy C-mean clustering algorithm

Under the wave of informationization, major universities have established their own experimental teaching platforms to facilitate student learning and improve teaching efficiency. Analyzing the large amount of learning data on the platform to help improve teaching design is a trend in the context of the big data era. Time series is an ancient subject, and by observing things, recording data according to the law of time, and analyzing the data, it is the analysis of time series data. The platform data contains characteristics of multiple dimensions such as college, major, gender, study time point, grade, etc. Considering the complexity of the factors affecting the grades, the method such as dividing the students of a certain college into one category cannot be simply used. In this paper, the fuzzy clustering algorithm is first used to analyze the clustering, and the time series data is further analyzed and processed after clustering.

2.3.1

Description of the fuzzy C-mean clustering algorithm

Fuzzy C-mean clustering algorithm has been developed by experts and scholars and is widely used in various fields such as agriculture, medicine, imaging, astronomy and so on.

The purpose of fuzzy C mean clustering algorithm is to make the objective function: (1) $J_{f c m} (U, V) = \sum_{i = 1}^{c} \sum_{j = 1}^{n} u_{i j}^{m} d_{i j}^{2}$

Minimized and satisfies $\sum_{i = 1}^{c} u_{i j} = 1 (j = 1, 2, \dots n)$ , n is the number of samples to be clustered, $\sum_{j = 1}^{n} u_{i j} \geq 0$ and u_ij ≥ 0, 1 ≤ i ≤ c, 1 ≤ j ≤ n. Where m is the fuzzy coefficient $(m > 1)$ , u_ij are the elements in the fuzzy matrix U representing the jth sample x_j belonging to the ith class with u_ij affiliation. $V_{k \times c} = (v_{1}, v_{2}, \dots, v_{c})$ , v_i is the ith clustering center vector. $d_{i j} = ‖ x_{j} - v_{i} ‖$ is the Euclidean distance between the sample point and the clustering center.

Steps for calculating the fuzzy c-means algorithm:

Step1: Initialization. Set the number of clusters c(1 < c < n), fuzzy coefficients m(m ≥ 1), convergence accuracy ε > 0 (ε is the iteration termination condition), and initialize the affiliation matrix U with random numbers.

Step2: Calculate the clustering center. (2) $v_{i} = \sum_{j = 1}^{n} u_{i j}^{m} x_{j} / \sum_{j = 1}^{n} u_{i j}^{m}, 1 \leq i \leq c$

Step3: Update the affiliation matrix. (3) $u_{i j} {\begin{array}{l} 1 / \sum_{k = 1}^{c} {(\frac{d_{i j}}{d_{k j}})}^{\frac{2}{m - 1}} & H_{j} = \emptyset \\ 1 / | H_{j} | & H_{j} \neq ⊘, i \in H_{j} \\ 0 & H_{j} \neq ⊘, i \notin H_{j} \end{array}$

where $H_{j} = {(i, j) | x_{j} = v_{i}, i = 1, 2, \dots, c}$

Step4: Repeat Step2 and Step3 and end when the affiliation matrix U stops changing. (4) $‖ U^{(s)} - U^{(s + 1)} ‖ \leq ε, s \geq 1$

2.3.2

Design of DTW-based fuzzy C-mean clustering algorithm

The d_ij in the formula of the traditional fuzzy c-mean algorithm represents the Euclidean distance between the sample points and the clustering center. DTW is used to calculate the similarity distance, and a fuzzy C-means clustering algorithm based on the DTW similarity measure is designed. Then the objective function of DTW-FCM is (5) $J_{f c m} (U, V) = \sum_{i = 1}^{c} \sum_{j = 1}^{n} u_{i j}^{m} d_{D T w}^{2} (x_{j}^{t}, v_{i}^{t})$

where $d_{D T W} (x_{j}^{t}, v_{i}^{t})$ is the DTW similarity distance between the time series $x_{j}^{t}$ corresponding to sample point x_j and the time series $v_{i}^{t}$ corresponding to clustering center v_i. During the experiment calculation d_DTW is based on dynamic programming to construct the distance matrix d[i][j] of sequence a and sequence b, where d[i][j] represents the square of the similarity distance between sequences a[0:i] and b[0:j] as shown in equation (6): (6) $d [i] [j] = {\begin{matrix} {(a [0] - b [0])}^{2} & i = 0, j = 0 \\ {(a [0] - b [j])}^{2} + d [0] [j - 1] & i = 0, j > 0 \\ {(a [i] - b [0])}^{2} + d [i - 1] [0] & j = 0, i > 0 \\ \begin{matrix} {(a [i] - b [j])}^{2} \\ + \min (d [i - 1] [j], d [i] [j - 1], d [i - 1] [j - 1]) \end{matrix} & i, j > 0 \end{matrix}$

The process of fuzzy C-mean clustering algorithm based on DTW is as follows:

Step1: Initialization. Set the number of clusters c(1 < c < n), fuzzy coefficient m(m ≥ 1), convergence accuracy ε > 0 (ε is the iteration termination condition), and initialize the affiliation matrix U with random numbers.

Step2: Calculate the clustering center. (7) $v_{i} = \sum_{j = 1}^{n} u_{i j}^{m} x_{j} / \sum_{j = 1}^{n} u_{i j}^{m}, 1 \leq i \leq c$

Step3: Update the affiliation matrix. (8) $u_{i j} {\begin{matrix} 1 / \sum_{k = 1}^{c} {(\frac{d_{D T W} (x_{j}^{t}, v_{i}^{t})}{d_{D T w} (x_{j}^{t}, v_{k}^{t})})}^{\frac{2}{m - 1}} & H_{j} = ⊘ \\ 1 / | H_{j} | & H_{j} \neq ⊘, i \in H_{j} \\ 0 & H_{j} \neq ⊘, i \notin H_{j} \end{matrix}$

where $H_{j} = {(i, j) | x_{j}^{t} = v_{i}^{t}, i = 1, 2, \dots, c}$ , when H_j = ⊘, the time series $x_{p}^{t}$ corresponding to the most neighboring v_i sample point x_p is selected instead of $v_{i}^{t}, x_{k}^{t}$ for $v_{k}^{t} (k = 1, 2, \dots, c)$ , i.e., $d_{D T w} (x_{j}^{t}, x_{p}^{t})$ and $d_{D T W} (x_{j}^{t}, x_{k}^{t})$ are used instead of $d_{D T w} (x_{j}^{t}, v_{i}^{t})$ and $d_{D T W} (x_{j}^{t}, v_{k}^{t})$ , respectively, in Eqs. Where the closest neighboring v_i sample points are determined using Euclidean distance.

Step4: Repeat Step2 and Step3, ending when the affiliation matrix U stops changing (9) $‖ U^{(s)} - U^{(s + 1)} ‖ \leq ε, s \geq 1$

3

Data analysis of the all-media experimental platform based on DTW-FCM

The all-media experimental platform constructed in this paper is provided for news learners to use, and the data of learners’ learning records are analyzed by using DTW-FCM, and corresponding teaching suggestions are given according to the analysis results. The following section will give a specific description of the data analysis.

3.1

Analysis of Learner Types

There are 42,159 learners’ learning records data in the all-media experimental platform used in this paper, of which 5,280 are Chinese learners. We classify the learners into four categories according to the categorization method in the platform data, namely, registered learners, browsing learners, exploring learners, and certificate-acquiring learners, and we can also derive the proportion of each category of learners through the records in the dataset.

Figure 4 represents the proportion of learners in each category. According to Fig. 4, it can be seen that the proportion of learners who can obtain certificates in platform learning is 2.9%, and the sum of the proportions of explorers and those who have obtained certificates is only 6.5%, while the two proportions of platform learners in China are 1.3% and 3.5% respectively. From this data analysis, it can be seen that there are fewer explorers in platform learning, the proportion of learners who can obtain certificates at the end of the course is too low, and more than half of the learners are browsers, who only read part of the course content after registering for the course.

3.2

Experimental Analysis and Results of Learning Behavioral Characteristics

The data of the all-media experimental platform used in this paper contains typical attributes of learning behaviors such as “days of learning interactions, number of video learning times, number of visits to the forum, number of courseware learning times, number of learning interactions,” etc. This part of the paper carries out a statistical analysis of these five attributes, and the specific information is shown in Table 1. At the same time, in order to analyze more intuitively the correlation between various types of learning behaviors and the final effect of e-learning, we selected five attributes of learning behaviors and conducted a preliminary analysis by means of a scatter plot (see Figure 5).

Table 1.

Statistics of relevant characteristics

Statistic	Minimum value	Maximum value	Mean value	Standard deviation
Behavior characteristic	Minimum value	Maximum value	Mean value	Standard deviation
Learning interaction days	1	208	4.5	10.6
Learning interaction frequency	1	197789	296.9	1252.8
Number of chapters learned	1	46	2.1	3.9
Number of forum speakers	0	20	0.04	0.22
Number of video plays	1	98520	35	264.6

As can be seen from Table 1 and Figure 5, for the attribute of the number of learning interactions, except for some learners who reached the extremely high value of 197,789 interactions, the relative number of interactions of the majority of learners is relatively low, and the difference between the number of interactions of different learners does not reflect a more obvious gap in the learning achievement.

In the attribute of learning interaction days, the maximum value is 208, the average value is 4.5, and according to the distribution trend in the scatter plot, it can be seen that there is a positive correlation between the number of course interaction days and the course grade, but this trend is not very clear, probably due to the limitation of the other characteristic attributes.

The maximum value of the number of video playbacks is 98520, the average value is 35, and there is no obvious correlation trend between the number of video playbacks and the course grade according to the distribution trend in the scatter plot. The maximum value of the number of chapters studied is 46, and the mean value is 2.1. There is a relatively clear positive correlation trend between the number of chapters studied and the grade, i.e. the more chapters studied, the higher the grade is likely to be. The maximum value of the number of forum speeches is 20, and the average value is 0.04. It can be seen that the number of forum speeches and discussions of the learners is extremely low, and the academic performance is more densely distributed on both sides of the scatterplot, which does not show a clear linear trend, but there is a tendency of polarization in the academic performance of the learners.

3.3

Behavioral characteristics of each category after cluster analysis

The normalized data was clustered with a number of clusters from 2 to 9, and the scores for clusters 2 to 9 were calculated separately using the variance ratio criterion method. The results of the score visualization are shown in Figure 6, where 2-9 is the number of clusters and 500-1100 is the corresponding score for each number.

From Figure 6, it can be seen that the highest score is divided into 6 categories, i.e., it is most appropriate to divide the data of students’ online learning behaviors into 6 categories. The data on students’ online learning behaviors were divided into 6 categories (see Table 2 for details), and the average score for each category was calculated. Because the data were standardized to the values in the 0-1 interval, the calculation of the mean value of the values in this interval does not reflect the original data, so the mean value of each category after clustering should be reverse standardized.

Table 2.

Learning categories generated by cluster analysis

Category	Number of people	Average proximity/day	Frequency mean/times	Duration Average/minute
1	375	2.06	120.76	960.43
2	570	7.09	110.45	950.26
3	270	7.68	70.53	680.72
4	195	40.21	60.15	550.41
5	60	15.83	50.81	420.86
6	30	80.52	20.57	210.29
Total	1500	25.57	72.21	528.83

Combining the results of cluster analysis and the proportion of learners in each category can help teachers and platform administrators to develop a reasonable teaching strategy, the following behavioral characteristics of each category to do a specific analysis. 1)

Focused learners

This type of learners has a high learning time and frequency, and the last online learning time is quite short from the current time. They are usually very focused on learning, and will continue to invest a lot of time and energy in the learning process. In order to keep them focused and achieve better learning results, teachers can provide them with more in-depth learning materials, challenging learning tasks, and more feedback to stimulate their interest and motivation.

2)

Continuous Learners

This type of learner has a higher length and frequency of learning, but their latest online learning time is shorter than their current time. They usually have certain learning goals and will try their best to maintain the continuity of learning. In order to help them achieve their learning goals, teachers can provide them with learning tasks with clear goals and feasible plans, and encourage them to make learning plans and record their learning progress all the time to maintain learning continuity.

3)

Persistent Learners

The length and frequency of learning of this kind of learners are high, and the latest online learning time is not too long from the current time. They usually learn online during their free time or when they need to learn. In order to help them make better use of their learning time, teachers can provide them with diversified learning resources and flexible learning modes, and encourage them to make reasonable learning plans and plan their learning time to improve their learning efficiency.

4)

Suspended learners

This kind of learners used to have high learning hours and frequency, but are currently in a state of suspension, with the latest online learning time being a long time from the current time. They usually interrupt their learning temporarily for some reason and are likely to continue learning. In order to help them restart learning, teachers can provide them with some help and support, such as reassigning learning tasks, providing personalized learning plans, and providing psychological support, etc., so as to facilitate their re-engagement in learning.

5)

Occasional learners

This kind of learners have a lower length and frequency of learning, and their latest online learning time is also longer from the current time. They usually have certain learning needs, but may be affected by other factors that prevent them from continuing online learning. In order to help them learn better, teachers can provide them with flexible learning resources and learning modes, and try to understand their learning needs and limitations, and provide targeted support and assistance to help them learn better.

6)

Abandoned learners

The length and frequency of learning for this type of learners are extremely low, and the latest online learning time is very long from the current time. They usually used to engage in online learning but have abandoned it for some reasons. In order to encourage them to re-engage in learning, teachers can provide them with some stimulation and encouragement, e.g. providing some interesting learning tasks or challenges, providing some practical learning materials, providing personalized learning plans, etc., so as to re-stimulate their interest and motivation in learning.

4

Conclusion

This paper studies the construction and data analysis of an all-media experimental platform for teaching journalism and communication needs. Based on the fuzzy C-mean clustering algorithm of DTW, the data related to journalism students’ use of the all-media experimental platform constructed in this paper are statistically analyzed. The proportion of learners in the platform learning who can obtain the certificate is 2.9%, and the sum of the proportion of explorers and those who obtain the certificate is only 6.5%, which is a low overall proportion. There is a certain positive correlation trend between typical learning behavior-related attributes such as “number of days of learning interaction, number of video learning, number of forum visits, number of courseware learning, number of learning interactions” and learning achievement. Through cluster analysis, the student behavior categories are divided into 6 categories, among which the proportion of continuous learners reaches 38%, the proportion of focused learners accounts for 25%, and the proportion of learners in other categories ranges from 2% to 18%. There are differences in the goals and ways of learning using the all-media experimental platform among learners in each category, which require teachers to provide targeted teaching assistance.

Using the all-media experimental platform to assist students’ learning and teachers’ teaching, we always pay attention to the learning needs behind the students’ behavioral data and provide corresponding help. This has greater benefits for enhancing the learning confidence and enthusiasm of journalism and communication students, as well as improving their learning ability.

Idioma:: Inglés

Calendario de la edición:: 1 veces al año
Temas de la revista:: Ciencias de la vida, Ciencias de la vida, otros, Matemáticas, Matemáticas aplicadas, Matemáticas generales, Física, Física, otros

RSS Feed de revista

The construction of a multi-platform system for teaching journalism and communication based on big data technology

Yun Liu

Publicado en línea: 24 mar 2025

Recibido: 04 nov 2024

Aceptado: 06 feb 2025

DOI: https://doi.org/10.2478/amns-2025-0782

Palabras claveNews communication teaching, All-media experimental platform, DTW, Fuzzy C-mean clustering, Behavioral characteristics

© 2025 Yun Liu, published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Palabras clave
News communication teaching, All-media experimental platform, DTW, Fuzzy C-mean clustering, Behavioral characteristics