View in EDS HTML Full Text PDF Full Text

The Dynamics of Social Performance and Cognitive Depth between Students and Teacher in Online Discussion Forums with the SNA and LDA Approach

Saved in:
Bibliographic Details
Title: The Dynamics of Social Performance and Cognitive Depth between Students and Teacher in Online Discussion Forums with the SNA and LDA Approach
Language: English
Authors: Wei Xu (ORCID 0000-0002-9042-213X), Yuhan Chen, Leying Yang
Source: Innovations in Education and Teaching International. 2025 62(1):135-151.
Availability: Routledge. Available from: Taylor & Francis, Ltd. 530 Walnut Street Suite 850, Philadelphia, PA 19106. Tel: 800-354-1420; Tel: 215-625-8900; Fax: 215-207-0050; Web site: http://www.tandf.co.uk/journals
Peer Reviewed: Y
Page Count: 17
Publication Date: 2025
Document Type: Journal Articles
Reports - Research
Education Level: Higher Education
Postsecondary Education
Descriptors: Cognitive Processes, Electronic Learning, Discussion (Teaching Technique), Computer Mediated Communication, Teacher Student Relationship, Student Role, Teacher Role, Differences, Prompting, Student Participation, Academic Achievement, Change, College Students, College Faculty, Social Networks, Graphic Arts
DOI: 10.1080/14703297.2023.2282155
ISSN: 1470-3297
1470-3300
Abstract: Online discussion forums are crucial educational tools that facilitate interaction between students and teachers. We created an online discussion forum, with the teacher serving as the moderator. 58 posts and 1,955 comments were collected. We analysed these data through Latent Dirichlet Allocation to determine word cooccurrence and extract key topics from the discussions. We examined how the topics and depth of the discussion changed over time and the roles that the students and teacher played in the forum by using time-series data. The results indicate that the teacher played a crucial role in the forum by initially serving as a moderator and then guiding or prompting students to discuss certain topics. Those who participated more actively achieved higher grades, and those that were passive had lower grades at the end Overall, the online discussion forum enhanced the course by helping students understand the core materials and topics in the course.
Abstractor: As Provided
Entry Date: 2025
Accession Number: EJ1458056
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwFT2Pou4yapvrHxVmVRwRVrAAAA4zCB4AYJKoZIhvcNAQcGoIHSMIHPAgEAMIHJBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDP2v_Ffypsym1FpJ7wIBEICBmxMwicrxkzTOxR1xTO_l7rP4aNZXJJfAEY_UjEb8FrQpUvBbO-t8UmONOHIo9GuD8imQlWufQbq-7Qxwit0-Ctl2ypo7tOGx54OmELtAcvea3xRMM3o5CAiwyNMEEeFNGRLbT98XNEp0zS1YlAe-739JFVLFj8STcpzQ7mn63wGqt7X8JkVFzalF4dxWIJMOZCmDP1NuAXQACLcn
Text:
  Availability: 1
  Value: <anid>AN0182296574;hzj01feb.25;2025Jan21.00:37;v2.2.500</anid> <title id="AN0182296574-1">The dynamics of social performance and cognitive depth between students and teacher in online discussion forums with the SNA and LDA approach </title> <p>Online discussion forums are crucial educational tools that facilitate interaction between students and teachers. We created an online discussion forum, with the teacher serving as the moderator. 58 posts and 1,955 comments were collected. We analysed these data through Latent Dirichlet Allocation to determine word cooccurrence and extract key topics from the discussions. We examined how the topics and depth of the discussion changed over time and the roles that the students and teacher played in the forum by using time-series data. The results indicate that the teacher played a crucial role in the forum by initially serving as a moderator and then guiding or prompting students to discuss certain topics. Those who participated more actively achieved higher grades, and those that were passive had lower grades at the end Overall, the online discussion forum enhanced the course by helping students understand the core materials and topics in the course.</p> <p>Keywords: Higher education; online discussion forum; Latent Dirichlet Allocation; time-series data</p> <hd id="AN0182296574-2">Introduction</hd> <p>Online learning has become a widely recognised and implemented learning method in the education sector, particularly during the new coronavirus outbreak, providing a strong guarantee for the orderly growth of education and instruction (Wu et al., [<reflink idref="bib26" id="ref1">26</reflink>]). The benefits of online learning have not been overlooked in the post-epidemic era, the online and offline blended teaching mode has gradually become a new paradigm of education. In higher education, the online-offline blended teaching can provide a more flexible and effective learning situation. As an important platform for online learning interactions, online discussion forums support meaningful discussions between students and teachers (Wang, [<reflink idref="bib22" id="ref2">22</reflink>]), allow students to gradually establish a learning community through knowledge construction. However, there is still uncertainty in online discussion forums in actual teaching, which decreases the quality and efficiency of online learning interactions (Liu, [<reflink idref="bib13" id="ref3">13</reflink>]). As a result, it is critical to clarify the internal interaction logic of online learning interactions and increase their quality.</p> <p>Researchers have explored patterns in large quantities of data on online discussions in education (Wang, [<reflink idref="bib23" id="ref4">23</reflink>]) to improve their quality. However, the majority of studies have ignored the changing patterns of cognitive and social levels (Galikyan et al., [<reflink idref="bib6" id="ref5">6</reflink>]). The Latent Dirichlet Allocation (LDA) topic model and Social Network Analysis (SNA) were used in this study to investigate topic changes and the dynamic nature of interaction in online discussion forums. It also provides recommendations to help teachers use online discussion forums in online teaching and improve knowledge construction.</p> <hd id="AN0182296574-3">Literature review</hd> <p></p> <hd id="AN0182296574-4">Online discussion forum</hd> <p>Online discussion forums have been extensively employed in blended learning situations, particularly in the field of higher education, where strong social connections are seen as being crucial to improving student achievement (Chen et al., [<reflink idref="bib1" id="ref6">1</reflink>]). Online discussion forums currently struggle with a lack of meaningful participation (Galikyan et al., [<reflink idref="bib6" id="ref7">6</reflink>]), a lack of debate focus in the topic matter, tardy teacher comments (Zhang et al., [<reflink idref="bib31" id="ref8">31</reflink>]), and a need to deepen the discussion topics (Gao et al., [<reflink idref="bib7" id="ref9">7</reflink>]). In order to enhance the pedagogy of online discussion forums, studies have sought to mine the interaction patterns of forums using learning analytics. For instance, Wong et al. ([<reflink idref="bib25" id="ref10">25</reflink>]) visualised the relationship between online interactions and contextual topics using Topic Modelling and SNA. Han et al. ([<reflink idref="bib8" id="ref11">8</reflink>]) investigated the cognitive-emotional traits of learners in various learning styles using deep learning and SNA. Numerous have also examined the effects of online discussion forums on cognitive aspects (Warren & Paulus, [<reflink idref="bib24" id="ref12">24</reflink>]) and learning performance (Liu et al., [<reflink idref="bib14" id="ref13">14</reflink>]) in order to better understand how they affect learning outcomes. Positive social behaviours in online discussion forums have been shown to affect students' final grades (He et al., [<reflink idref="bib9" id="ref14">9</reflink>]). Additionally, there is growing interest in the use of teacher feedback as an intervention in online discussion forums. Costley ([<reflink idref="bib4" id="ref15">4</reflink>]) assert that while teachers' direct instruction improves students' critical thinking, facilitating discourse can enhance students' social presence to support online interactions among students. Wu et al. ([<reflink idref="bib26" id="ref16">26</reflink>]) investigated the impact of various teacher feedback techniques on students' participation in online conversations.</p> <hd id="AN0182296574-5">SNA of online learning communities</hd> <p>SNA is a method for detecting network structures and is widely used to explore interactions between node networks (Wu & Nian, [<reflink idref="bib27" id="ref17">27</reflink>]). Online discussion forum is a very typical application scenario of SNA, which forms an interactive network with the Posting and reply behaviours of teachers and students as well as students (Jiang & Wang, [<reflink idref="bib10" id="ref18">10</reflink>]). Meanwhile, whether it is based on communication or association, affects students' behaviours. For this reason, researchers have used SNA indicators to evaluate online learning communities and students' participation.</p> <p>Most studies applying SNA to online learning communities have used either individual- and network-level indicators to evaluate interaction networks and the roles of learners and teachers (Ouyang & Scharber, [<reflink idref="bib17" id="ref19">17</reflink>] or have combined SNA with analysis of interactivity to evaluate performance, emotions, motivation, and other aspects related to students. Xu and Chen ([<reflink idref="bib29" id="ref20">29</reflink>]) combined content analysis with SNA to explore students' social behaviour and knowledge construction during the process of forming a learning community. Their results revealed that changes in students' and teachers' roles can promote knowledge construction. Through text mining and SNA, Xie et al. ([<reflink idref="bib28" id="ref21">28</reflink>]) explored leadership among students in a forum for an online course and developed an educational intervention on the basis of the results. Han et al. ([<reflink idref="bib8" id="ref22">8</reflink>]), combined deep learning and SNA to identify cognitive-emotional patterns related to interaction on a platform for Massive Open Online Courses (Han et al., [<reflink idref="bib8" id="ref23">8</reflink>]). Studies have demonstrated that SNA is an effective method for analysing interactive networks in computer-based cooperative and exploratory education (Wu & Nian, [<reflink idref="bib27" id="ref24">27</reflink>]).</p> <hd id="AN0182296574-6">Clustering topics in text from online interactions through LDA</hd> <p>The learning data generated in the interactive process of online learning not only includes explicit interactive behaviour, but also includes interactive text data (Chen et al., [<reflink idref="bib3" id="ref25">3</reflink>]). With large quantities of text data, traditional content analysis is time consuming and laborious; to overcome this limitation, automated topic modelling can be applied for the textual analysis of data on online interactions. LDA, a classical text-clustering method based on the Bayesian model, includes three layers: documents, topics, and words (Zhang et al., [<reflink idref="bib30" id="ref26">30</reflink>]), and it can be used to generate document – topic and topic – word probability distributions and thus identify and extract potential topic hotspots (Shi, [<reflink idref="bib21" id="ref27">21</reflink>]).</p> <p>However, because LDA exhibits low accuracy in the clustering of short passages of text, researchers have optimised LDA through various methods (Chen & Wu, [<reflink idref="bib2" id="ref28">2</reflink>]). For example, Rosen-Zvi et al. ([<reflink idref="bib19" id="ref29">19</reflink>]) refined the application situation of topic model and proposed the concept of document-author topic distribution for the first time. Ozyurt and Ali Akcayol ([<reflink idref="bib18" id="ref30">18</reflink>]) suggested an emotion-based method for sentiment analysis of short text passages. Numerous studies have linked topic modelling to learning behaviour (Liu, [<reflink idref="bib13" id="ref31">13</reflink>]) and learning performance (Zhang et al., [<reflink idref="bib30" id="ref32">30</reflink>]), and try to tap into the differences in topics for different user groups (Li et al., [<reflink idref="bib15" id="ref33">15</reflink>]).Thus, LDA has strong potential for analyses of textual data on interactions because it enables automation and visualisation.</p> <hd id="AN0182296574-7">Research question</hd> <p>Online discussion forums facilitate the creation of collective knowledge because they enable individual and collective learning through interaction among members (Lin et al., [<reflink idref="bib12" id="ref34">12</reflink>]). This study focused on how teacher engagement impact student engagement and analysed social and cognitive changes among students participating in the online discussion forum. Specifically, this study addressed the following research questions:</p> <p></p> <ulist> <item> How does the social network of an online discussion forum change over time?</item> <p></p> <item> How do teachers socialise with students of different academic achievement over time?</item> <p></p> <item> What are the key topics of discussion in such forums, and how do they change over time?</item> </ulist> <hd id="AN0182296574-8">Method</hd> <p></p> <hd id="AN0182296574-9">Method</hd> <p>A mixed research methodology was employed in this study to assess and retrieve the data from online discussion forums. To determine how evaluate the dynamic changes of the online discussion forum, quantitative SNA measurements and LDA topic modelling were used. To better identify the changes in students' cognitive levels throughout the collaborative process and because the issue of LDA necessitates human generalisation, the qualitative approach is complemented by developing a distinctive dimensional attribute coding system.</p> <hd id="AN0182296574-10">Participants and context</hd> <p>This study recruited 44 students studying educational technology (average age = 19 years) and 1 teacher from a professional elective course on graphic design in the educational technology programme featuring both offline and asynchronous online teaching. At the beginning of the course, the teacher created an online discussion forum on the Learning Connect platform for the students as an extension of the class. The teacher also encouraged the students to participate in the online discussions.</p> <hd id="AN0182296574-11">Study design and procedure</hd> <p>Figure 1 presents the research processes. In this study, a 6-week graphic design course was described. Before each lesson, the teacher would post pre-assigned reading materials to the online discussion forum so that students can discuss topics in advance. During the course, teacher employed the lecture and the task-driven approach. She also used the online discussion forum to post ill-structured problems related to the course content. After the course, students should finish the work homework for ordinary grades. In the meantime, students can share information, establish interactions and ask for and give help in the online discussion forum. On the other hand, the teacher participated by liking, replying, giving direct instructions, and posting topics related to student confusion. The final course grades of the students as well as interaction data from the online discussion forums were gathered as data sources for learning analytics. This study utilises LDA and SNA to explore the dynamics of socialisation and cognition in online discussion forums and further extends the results to students' grades.</p> <p>Graph: Figure 1. Research processes.</p> <hd id="AN0182296574-12">Measurements</hd> <p>The Learning Connect platform was used to collect data for this study, which included the interaction time and content (58 posts and 1,965 comments) in the online discussion forum, as well as the students' final course grades.</p> <p>To increase the objectivity of LDA topic condensation, the study built a coding system based on the teaching objectives and knowledge points and retrieved unified attribute dimensions for the theme feature words (Chen et al., [<reflink idref="bib3" id="ref35">3</reflink>]). As shown in Table 1, the coding framework includes eight knowledge attributes and is divided into three cognitive levels: shallow, medium, and deep. From shallow to deep level, these attributes are thematic relevance (SL1), basic operation (SL2), visual effect (SL3), professional term (SL4), composition of a picture (ML1), colour matching (ML2), design skills (ML3), attitude and literacy (DL).</p> <p>Table 1. Feature dimension attribute encoding framework.</p> <p> <ephtml> <table><thead><tr><td /><td>Level 1 dimension</td><td>attribute</td><td>Attribute coding</td><td>Code interpretation</td><td>Examples of feature words</td></tr></thead><tbody><tr><td>Shallow</td><td>SL</td><td>Thematic relevance</td><td>SL1</td><td>Feature words that are consistent with the topic words of the discussion</td><td>Major, replace, work</td></tr><tr><td /><td /><td>Basic operation</td><td>SL2</td><td>The operations involved in the design process</td><td>alt、ctrl、copy</td></tr><tr><td /><td /><td>Visual effect</td><td>SL3</td><td>The visual feeling brought by the work</td><td>Visual effects, impact, exaggeration</td></tr><tr><td /><td /><td>Professional term</td><td>SL4</td><td>Vocabulary specific to graphic design courses</td><td>Color levels, closure, filters</td></tr><tr><td>Deep</td><td>ML</td><td>Composition of a picture</td><td>ML1</td><td>The operation and arrangement of the position on the picture</td><td>Structure, gravity, orientation</td></tr><tr><td /><td /><td>Color matching</td><td>ML2</td><td>Color collocation and planning</td><td>Mix colour, tone</td></tr><tr><td /><td /><td>Design skills</td><td>ML3</td><td>The skills used in the design and production of the work</td><td>Leave white space and reconcile</td></tr><tr><td /><td>DL</td><td>Attitude and literacy</td><td>DL</td><td>Emotional attitude and innovation, information literacy</td><td>Copyright, commercial, claims</td></tr></tbody></table> </ephtml> </p> <p>1 <emph>Analysis</emph>.</p> <hd id="AN0182296574-13">Analysis</hd> <p>The 6-week course was separated into three periods, and students' grades were sorted into four groups: high, above average, average, and below average.</p> <hd id="AN0182296574-14">Data preprocessing</hd> <p>First, duplicate posts were eliminated in order to acquire 58 posts and 1955 legitimate comments as the source data for SNA. Second, to collect 1967 valid sentences as the initial data set for LDA, invalid posts that only contained picture comments were eliminated. The nonsense words were further cleaned by filtering the parts of speech and stopping words, replacing synonyms and constructing a custom dictionary, and then the text was vectorised.</p> <hd id="AN0182296574-15">SNA for network and node levels and analysis of correlations between students' social charact...</hd> <p>The Igraph and visNetwork package in R language was used to detect SNA measures. The Corrplot packages were used to get the indicators of the correlation between students' student grades and social characteristics.</p> <hd id="AN0182296574-16">LDA topic clustering and feature attribute division</hd> <p>Python's Sklearn library was used for LDA to extract Topic – word and document – topic distributions. According to the eight knowledge attributes in Table 1, the topic feature words' feature attributes were annotated. The cumulative probability of the feature words falling under various feature attributes and the probability of each topic feature word involving various attributes were then calculated. The topic's dominant attribute is determined by which attribute has the highest likelihood of appearing in the results, and the topic's dominant dimension is determined by which dimension this attribute corresponds to.</p> <hd id="AN0182296574-17">Network analysis of cooccurrence of feature words and evolution of key topics over time</hd> <p>The co-occurrence network analysis was carried out for the topic feature words with the same dominant attribute, and the line chart and boxplot of the frequency changes of each topic in the three periods were drawn, so as to explore the internal connection and dynamic development characteristics of topic hot spots.</p> <hd id="AN0182296574-18">Results</hd> <p></p> <hd id="AN0182296574-19">The evolution of learning communities: Changes in the social network over time</hd> <p>Figure 2 presents the changes in the social network in each of the three periods. The size of the nodes is related to centrality, the thickness of the lines represents the weight, and the colours indicate roles and final grades; green represents teachers, and all other colours represent students. Between the first and third periods, there were fewer isolated nodes and a noticeable rise in node size, which showed that there were more participants and that the discussion was more active. The students with the highest final grades (red and orange nodes) exhibited the high levels of participation in each period, whereas the students with low grades (yellow and blue nodes) exhibited the low levels of participation or did not participate in the discussion. From the network level, the network connections strengthened across the three periods.</p> <p>Graph: Figure 2. Three periods of social network of online discussion forum.</p> <p>Through the descriptive statistics of the key data of the three-period network and the measurement at node level and network level, the dynamic changes of the network in each index are further understood. Between the first and third periods, the number of comments (195-271-1,<reflink idref="bib438" id="ref36">438</reflink>) and posts (<reflink idref="bib2" id="ref37">2-8-36</reflink>) from the students increased significantly. The average degree (2-3.82-34.31), density (0.145–0.760), and global clustering coefficient (0.054–0.841) of the nodes from low to high indicated that the discussion network became denser as the course progressed. The small average path length (1.844–1.265) indicated that the relationships between students in the network depended on one person on average, and that the network had high connectivity. The large increase in the number of mutual (13–236) and asymmetric (168–965) duals and the low to medium reciprocity (0.072–0.197) indicated that the students' sense of reciprocity increased over time. The number of nodes in strongly and weakly connected components increased, which indicates an increase in closeness of propagation in the online discussion forum and the students gradually developed cooperative relationships.</p> <hd id="AN0182296574-20">Changing roles of teachers and students: SNA for changes in the roles of students and teacher...</hd> <p>To gain further insights into the roles of the teacher and students, a heat map of individual-level SNA indicators in descending order in each period was created (Figure 3).</p> <p>Graph: Figure 3. SNA measures of each participant in each period.</p> <p>In the first period, the teacher led the students through building the community. The teacher played a central role and had a strong ability to control teaching resources. The students relied on posts created by teachers and were inclined to comment on the teacher's posts. This observation indicated the teacher's high outdegree (<reflink idref="bib12" id="ref38">12</reflink>), indegree (<reflink idref="bib178" id="ref39">178</reflink>) and betweenness centrality (0.1929) and the only two students' posts were uploaded. Six students played the role of observers and did not participate in the discussion (all outdegree were 0).</p> <p>In the second period, all students participated, and the teacher played the role of facilitator, only posting two posts and responding to the students more, as indicated by the high outdegree (<reflink idref="bib29" id="ref40">29</reflink>). However, the teacher exhibited a high indegree (<reflink idref="bib248" id="ref41">248</reflink>), indicating that the students were enthusiastic about the teacher's posts and relied on them. Some students with high grades, such as S3, S22, and S25, began to create new topics and influence the discussion.</p> <p>In the third period, the overall student indegree, outdegree, and degree increased substantially. The teacher only played the role of coordinator and provided guidance when the students were confused or disagreed with each other. Unlike the prior two periods, students created new topics and replied to each other rather than relying on the teacher's posts. Students S27 and S2 had the highest betweenness centrality, indicating that they transmitted information throughout the network and played critical roles. Students S5, S21, and S1 demonstrated closeness centrality as well as fast information transfer speeds. Some students, such as S4 and S34, initiated discussions and relied less on others.</p> <p>The teacher played the roles of leader, facilitator, and coordinator in the first, second, and third periods, respectively. Over time, the students participated more in the discussion, and high achievers were more willing to post and comment, as indicated by the fact that the top of the Figure 3 consisted of mostly high achievers (red squares). Some students, who demonstrated extremely high outdegree at the period 3, such as S17 May 2001have started to comprehend the effect of discussion participation on their final marks.</p> <p>To examine the relationship between SNA measures and the students' performance, a pairwise correlation analysis was performed. There were significant positive correlations between some of the individual SNA measures in the three periods, but there was no significant correlation between the first two periods of the grades and the individual SNA measures, and only in the third period did the grades have a significant positive correlation with the outdegree (<emph>r</emph> = 0.616***), degree (<emph>r</emph> = 0.495**) and the eigenvector centrality (<emph>r</emph> = 0.582***).</p> <hd id="AN0182296574-21">Feature distribution of discussion topics: Analysis of feature word attributes</hd> <p>LDA was used to cluster topics in the forum and explore the effect of teacher's teaching intervention on the topics of discussion. On the basis of the local minimum of confusion in the training process, the optimal number of topics was six. Because of the long-tail effect of the feature words for each topic (Liang & Li, [<reflink idref="bib11" id="ref42">11</reflink>]), the first 15 feature words for each topic were extracted to obtain their corresponding probabilities and attributes (Tables 2 and 3).</p> <p>Table 2. Attributes and probabilities of feature words in topics 0–2.</p> <p> <ephtml> <table><thead><tr><td>Topic 0</td><td>Probability</td><td>Attribute</td><td>Topic 1</td><td>Probability</td><td>Attribute</td><td>Topic 2</td><td>Probability</td><td>Attribute</td></tr></thead><tbody><tr><td>company</td><td>0.034</td><td>DL</td><td>plagiarize</td><td>0.042</td><td>DL</td><td>alt</td><td>0.104</td><td>SL2</td></tr><tr><td>copyright</td><td>0.031</td><td>DL</td><td>direction</td><td>0.029</td><td>ML1</td><td>ctrl</td><td>0.080</td><td>SL2</td></tr><tr><td>infringement</td><td>0.028</td><td>DL</td><td>Painting style</td><td>0.029</td><td>SL4</td><td>layer</td><td>0.056</td><td>SL4</td></tr><tr><td>Commercial use</td><td>0.022</td><td>DL</td><td>white space</td><td>0.026</td><td>ML3</td><td>tool</td><td>0.050</td><td>SL4</td></tr><tr><td>law</td><td>0.020</td><td>DL</td><td>principle</td><td>0.025</td><td>SL4</td><td>shift</td><td>0.030</td><td>SL2</td></tr><tr><td>author</td><td>0.019</td><td>DL</td><td>harmony</td><td>0.024</td><td>ML3</td><td>adjust</td><td>0.023</td><td>SL2</td></tr><tr><td>authorization</td><td>0.018</td><td>DL</td><td>structure</td><td>0.023</td><td>ML1</td><td>color level</td><td>0.019</td><td>SL4</td></tr><tr><td>protect</td><td>0.017</td><td>DL</td><td>shape</td><td>0.022</td><td>ML3</td><td>option</td><td>0.014</td><td>SL4</td></tr><tr><td>product</td><td>0.015</td><td>SL1</td><td>gravity</td><td>0.020</td><td>ML1</td><td>mode</td><td>0.014</td><td>SL4</td></tr><tr><td>Character base</td><td>0.014</td><td>SL4</td><td>audiences</td><td>0.017</td><td>DL</td><td>show</td><td>0.013</td><td>SL2</td></tr><tr><td>literary property</td><td>0.011</td><td>DL</td><td>impact</td><td>0.016</td><td>SL3</td><td>selection</td><td>0.012</td><td>SL4</td></tr><tr><td>court</td><td>0.011</td><td>DL</td><td>background</td><td>0.016</td><td>ML1</td><td>brush</td><td>0.012</td><td>SL4</td></tr><tr><td>Claim for compensation</td><td>0.011</td><td>DL</td><td>Visual effect</td><td>0.014</td><td>SL3</td><td>nib</td><td>0.011</td><td>SL4</td></tr><tr><td>negotiate</td><td>0.010</td><td>DL</td><td>independent</td><td>0.013</td><td>DL</td><td>conversion</td><td>0.010</td><td>SL2</td></tr><tr><td>Right to use</td><td>0.010</td><td>DL</td><td>exaggerate</td><td>0.013</td><td>SL3</td><td>cancel</td><td>0.010</td><td>SL2</td></tr></tbody></table> </ephtml> </p> <p>Table 3. Attributes and probabilities of feature words in topics 3–5.</p> <p> <ephtml> <table><thead><tr><td>Topic 3</td><td>Probability</td><td>Attribute</td><td>Topic 4</td><td>Probability</td><td>Attribute</td><td>Topic 5</td><td>Probability</td><td>Attribute</td></tr></thead><tbody><tr><td>artistic</td><td>0.030</td><td>SL3</td><td>Artificial intelligence</td><td>0.037</td><td>DL</td><td>folder</td><td>0.060</td><td>SL4</td></tr><tr><td>principle</td><td>0.022</td><td>DL</td><td>replace</td><td>0.025</td><td>SL1</td><td>download</td><td>0.041</td><td>SL2</td></tr><tr><td>major</td><td>0.020</td><td>SL1</td><td>software</td><td>0.024</td><td>SL4</td><td>click</td><td>0.029</td><td>SL2</td></tr><tr><td>effect</td><td>0.019</td><td>SL3</td><td>work</td><td>0.024</td><td>SL1</td><td>copy</td><td>0.027</td><td>SL2</td></tr><tr><td>unify</td><td>0.017</td><td>SL1</td><td>mankind</td><td>0.018</td><td>SL1</td><td>color modulation</td><td>0.027</td><td>ML2</td></tr><tr><td>close</td><td>0.016</td><td>SL4</td><td>ability</td><td>0.015</td><td>DL</td><td>file</td><td>0.025</td><td>SL4</td></tr><tr><td>comprehend</td><td>0.016</td><td>DL</td><td>cartoon</td><td>0.015</td><td>SL4</td><td>install</td><td>0.023</td><td>SL2</td></tr><tr><td>public</td><td>0.016</td><td>DL</td><td>create</td><td>0.014</td><td>DL</td><td>gravity</td><td>0.020</td><td>ML1</td></tr><tr><td>style</td><td>0.015</td><td>SL4</td><td>function</td><td>0.011</td><td>SL4</td><td>decompression</td><td>0.019</td><td>SL2</td></tr><tr><td>trait</td><td>0.015</td><td>SL1</td><td>technology</td><td>0.011</td><td>SL4</td><td>filters</td><td>0.019</td><td>SL4</td></tr><tr><td>standard</td><td>0.015</td><td>SL1</td><td>Word soul</td><td>0.010</td><td>SL4</td><td>ins</td><td>0.019</td><td>SL4</td></tr><tr><td>hue</td><td>0.013</td><td>ML2</td><td>cartoon</td><td>0.009</td><td>SL4</td><td>plug</td><td>0.019</td><td>SL4</td></tr><tr><td>meaning</td><td>0.012</td><td>DL</td><td>inbetweening</td><td>0.009</td><td>SL4</td><td>curve</td><td>0.017</td><td>SL3</td></tr><tr><td>transmission</td><td>0.012</td><td>DL</td><td>major</td><td>0.009</td><td>SL1</td><td>paste</td><td>0.014</td><td>SL2</td></tr><tr><td>brush</td><td>0.012</td><td>SL4</td><td>art</td><td>0.008</td><td>SL4</td><td>illustrator</td><td>0.013</td><td>SL4</td></tr></tbody></table> </ephtml> </p> <p>The combination of feature dimension analysis and LDA topic clustering can provide guidance for interpreting posts. The probability distribution of each topic for each attribute was calculated (Table 4). 'Picture structure', 'colour collocation', and 'design skills' had the lowest total probability. The three attributes were in the medium-depth category, indicating that the theme in the whole discussion process had poor performance at the medium cognitive level. The dominant attribute of each topic was determined by identifying the attribute with the highest probability of contribution to the feature words for each topic. Differences in probability between attributes for each topic were observed. The dominant attributes of Topics 0 and 3 were 'attitude' and 'literacy', that of Topic 1 was 'Picture structure', that of Topic 2 was 'basic operation', and that of Topics 4 and 5 were 'professional terms'.</p> <p>Table 4. Feature distribution probability of each topic in different dimensions.</p> <p> <ephtml> <table><thead><tr><td /><td>Level 1 dimension</td><td>attribute</td><td>Topic 0</td><td>Topic 1</td><td>Topic 2</td><td>Topic 3</td><td>Topic 4</td><td>Topic 5</td><td>A total of probability</td></tr></thead><tbody><tr><td>Low level</td><td>SL</td><td>Thematic relevance （SL1）</td><td>0.015</td><td /><td /><td>0.067</td><td>0.076</td><td /><td>0.158</td></tr><tr><td /><td /><td>Basic operation （SL2）</td><td /><td /><td>0.27</td><td /><td /><td>0.153</td><td>0.423</td></tr><tr><td /><td /><td>Visual effect （SL3）</td><td /><td>0.043</td><td /><td>0.049</td><td /><td>0.017</td><td>0.109</td></tr><tr><td /><td /><td>Professional term （SL4）</td><td /><td>0.054</td><td>0.188</td><td>0.043</td><td>0.097</td><td>0.155</td><td>0.537</td></tr><tr><td>High level</td><td>ML</td><td>Composition of a picture （ML1）</td><td /><td>0.088</td><td /><td /><td /><td>0.020</td><td>0.108</td></tr><tr><td /><td /><td>Color matching （ML2）</td><td /><td /><td /><td>0.013</td><td /><td>0.027</td><td>0.04</td></tr><tr><td /><td /><td>Design skills （ML3）</td><td /><td>0.072</td><td /><td /><td /><td /><td>0.072</td></tr><tr><td /><td>DL</td><td>Attitude and literacy</td><td>0.242</td><td>0.072</td><td /><td>0.078</td><td>0.066</td><td /><td>0.458</td></tr></tbody></table> </ephtml> </p> <p>A cooccurrence network of subject feature words was examined to further investigate the relationships between feature words that share a dominant property. The greater the size of a node, the more connections it has with other nodes, and the thickness of the lines represents the edge weight.</p> <p>Figure 4 presents the cooccurrence network diagram of each topic. The discussion of professional terms centred on art terms related to graphic design and the corresponding terms related to software. The discussion involved artworks, forms of artistic expression, and artistic value.</p> <p>Graph: Figure 4. Cooccurrence network of high-frequency feature words for each topic.</p> <p>The cooccurrence of the 'gravity', 'structure', 'direction' and 'shape' nodes was significant for the 'composition of a picture' attribute in Topic 1 (Figure 4b), showing that the discussion of composition centred on these factors. Nodes with the 'composition of a picture' attribute cooccurred with nodes with other attributes, indicating that the composition discussion requires a certain amount of expertise.</p> <p>The dominant dimensions for Topics 0 and 3 were 'attitude' and 'literacy' (Figure 4c). 'Character base', 'copyright', 'author', 'literary property', 'business' and 'infringement' were the network's essential nodes. They regularly cooccurred, demonstrating that the students gradually formed an awareness of copyright via discussing artwork. The cooccurrence of the 'creator', 'transmission', and 'meaning' nodes revealed that the students' discussion of artworks focused on both the artworks' appearance and underlying meaning.</p> <p>Topic 2's dominant attribute was 'basic operations' (Figure 4d). The network connection and nodes with high degrees show that the students were willing to discuss Photoshop tools and shortcut keys.</p> <hd id="AN0182296574-22">Discuss the evolution of topic hotspots</hd> <p>Figure 5 shows the change in the frequency of thematic hotspots adjusted to the change in how the network as a whole interacts. The abscissa in Figure 5(a) depicts time, and the ordinate the frequency of conversation. The average heat value change for each topic is shown in a boxplot in Figure 5(b). The central line in the box denotes the average heat value, while the X-axis and Y-axis reflect time and the topic of discussion.</p> <p>Graph: Figure 5. The evolution of topic hotspots.</p> <p>The shift in social performance was consistent with an increase in the heat of all themes through time, and variations in heat were noted for each period. Topic 5 had the greatest average heat value and largest heat area in the first period. The line graph's curvature and the frequency of Topic 5 discussions both dramatically change in the later periods, suggesting that as the course material advanced, the online discussion forums began to focus more on terms related to graphic design and production. In the second period, the average heat values were low, and the heat region was small for all topics, indicating that the topics of discussion remained the same, that the discussion was not highly active, and that no particular topic was discussed. The third period saw a considerable change in the heat for Topics 0 and 3, as seen by the high mean values and sizeable heat regions for both topics. This is consistent with Figure 5(a), which shows that during the third period, these topics' curves all grew at various points. The Topics 1, 2, and 4 were consistently discussed in the third session, according to the mild heat mean and the extent of the heat region.</p> <p>The depth of discussion also evolved. Topics 4 and 5 were often discussed in the latter half of the first period and the beginning of the second period. Most discussions on the 'attitude' and 'literacy' themes were started by the teacher, but 'professional terms', which was in the shallow group, was the issue that predominated. The topic of Topic 1's composition was brought up frequently in the later part of the second session, indicating that the depth of the conversation grew over time. In the third session, Topics 0 and 3 for 'attitude' and 'literacy' had high average popularity scores, indicating more in-depth discussion.</p> <hd id="AN0182296574-23">Discussion and suggestions</hd> <p></p> <hd id="AN0182296574-24">Discussion</hd> <p></p> <hd id="AN0182296574-25">There are differences in social performance of students with different grades, which should g...</hd> <p>Differences in individual SNA measures in each period were observed among the grade groups. High- achieving students were eager to initiate discussions and play a central role in the network, including S2, S27, and S25. Low-achieving students were more inclined to reply and less likely to initiate discussions. The analysis of the correlation between SNA measures and final grades revealed that individual SNA measures were consistent predictors of grades, and that performance in the third period was positively related to partial SNA measures. This is consistent with the results of Saqr and Alamro ([<reflink idref="bib20" id="ref43">20</reflink>]), who observed a positive association between students' SNA measures and learning outcomes.</p> <p>Teachers should allow students to act on their subjective initiative and communicate offline with students who have low activity levels, and teachers provide targeted guidance and create opportunities for students to interact by replying to their comments, thereby strengthening their sense of presence in the network. Furthermore, teachers should exploit the network's core members' strength to motivate others, as these students' opinions may resonate with others.</p> <hd id="AN0182296574-26">The role strategy of teachers will promote the development of student learning community and...</hd> <p>The strategic change of teachers' roles can enhance online discussion forums and indirectly affect students' discussions. Studies have demonstrated that teachers acting as leaders in the early periods of courses are crucial to developing online discussion forums (Ouyang & Scharber, [<reflink idref="bib17" id="ref44">17</reflink>]). In the first period of the course, the teacher played the role of leader and had the highest betweenness centrality. The teacher posted nine posts related to the course to motivate them to interact in the online discussion forum. In the second period, the teacher became a facilitator, supervised the discussion, answered students' questions, and resolved conflicts (Wu & Nian, [<reflink idref="bib27" id="ref45">27</reflink>]). The students began to created their own topics. However, the students were still more likely to reply to the teacher's posts. In the third period, the teacher participated as a collaborator. The community had grown, and the teacher kept giving the students feedback to promote involvement and serve as a constant reminder (Martin et al., [<reflink idref="bib16" id="ref46">16</reflink>]).</p> <p>The effects of the teacher on students' behaviour and discussions indicated that the teacher played a key role. Therefore, teachers should play various roles in online discussion forums and adjust their roles depending on the progress of the discussion to facilitate knowledge exchange.</p> <hd id="AN0182296574-27">The cognitive depth of the discussion did not improve significantly with the advancement of t...</hd> <p>First of all, from a social network perspective, the high reciprocity over time was attributable to the frequency of discussion, which facilitated bidirectional interaction between nodes; the quality of core members' posts remained low, and many posts lacked depth. Secondly, in terms of topic classification, the three learning levels of shallow, medium and deep cover 3, 1 and 2 topics, respectively, with the overall discussion topic remaining primarily shallow. The cooccurrence network of feature words suggested that with the teacher's guidance, the students' discussions became more in depth. In terms of overall dynamics, the dimensions of the concerns that are discussed most frequently in the late first, late second, and third periods of the findings are shallow, medium, and deep, respectively. However, medium-depth discussion was lacking, and most discussions in the third period remained shallow.</p> <p>Teachers should introduce open-ended topics and provide support to facilitate deep discussion. In addition, teachers should enlist the support of core members, which would boost the average number of posts by 4 (Ozyurt & Ali Akcayol, [<reflink idref="bib18" id="ref47">18</reflink>]) and deepen the overall network activity. Teachers should also monitor the discussion, provide encouragement through comments, and indicate topics worth discussing to enrich the discourse.</p> <hd id="AN0182296574-28">Theme discussion is closely related to teachers' theme creation and attaches importance to pr...</hd> <p>The topic mainly followed the course's progress and was closely related to the topics posted by the teacher; hardly no text was unrelated to the course. The students tended to follow the teacher's model for posts because the teacher's posts were based on the progress of the course. In periods 1 to 3, students learn terminology, basic operations, and the creation of work, which correlate to the most interest topics in the forums for each period, such as professional term, basic operations, attitudes and literacy. The teacher monitored the students' attitude and literacy in the online discussion forum but paid little attention to the students' practical skills. The learning motivation is insufficient for the theme of comprehensive application, such as composition and design, which is particularly evident in the middle-order characteristic with the lowest total likelihood.</p> <p>Topics of discussion are closely related to teachers' guidance, as indicated by the lack of guidance in medium-depth topic discussion, resulting in poor performance in this regard. Therefore, teachers should identify the difficult topics in a course, broaden the scope of knowledge, enhance the interactive topic content step by step, and match the topics to the course progress to help students understand the material. Teachers should also macro-control the online discussion forum and prompt students to focus on points lacking in the discussion.</p> <hd id="AN0182296574-29">Limitations and implications for future research and practice</hd> <p>There are still some limitations to this study. Due to the limitations of the course schedule, the experimental period for this study only lasted 6 weeks, and there was no control group. It was challenging to monitor everything that was happening among the students in the online discussion forum and to provide them feedback in real time because there was only one teacher supervising the course. However, the temporal correlation between posts and comments was not significant.</p> <p>The future of education will alter drastically as a result of the quick development of AI technologies like ChatGPT. Three specific areas for future research include teaching implementation, learning tools, and teaching effect. First, in order to provide useful scaffolding for students and address the issue of tardy and incomplete instructor feedback due to its ability to generate logical responses based on complicated natural language models (Farrokhnia et al., [<reflink idref="bib5" id="ref48">5</reflink>]), ChatGPT can be utilised as a teaching aid to strengthen online discussion forums. Second, new analytics techniques are developed with new AI technologies to improve the feedback quality of learning analytics. Future research may consider textual algorithms to mine temporal changes in neighbouring postings to better define the correlations between pre- and post-discussion topics or to employ one- and two-mode SNA measures to understand how students relate to one another through topics. Third, the pedagogical benefits of online forums can be expanded to include higher-order skills and cognitive levels.</p> <hd id="AN0182296574-30">Conclusion</hd> <p>This study explores deeper into the interaction behaviours and topics of online discussion forums in higher education graphic design course. Technically, combining SNA with automated LDA topic models to mine interactive behaviours and this data, and using the unified feature dimension attribute coding framework to encode the feature words of LDA clustering, introduces a new way of thinking for learning and analysing online discussion forums. The link between SNA measurements and final course grades was also further investigated in this study. The results imply that teacher roles are essential for the development of learning communities, and that as learning communities are established, students develop a stronger sense of reciprocity, gradually wean themselves from teacher dependence, and experience cognitive shifts from shallow to deep topic discussion. Additionally, students with strong social skills had a higher chance of doing well on their final exams. Teachers should pay attention to planned topic creation and role transitions during the online discussion forums.</p> <hd id="AN0182296574-31">Disclosure statement</hd> <p>No potential conflict of interest was reported by the author(s).</p> <ref id="AN0182296574-32"> <title> References </title> <blist> <bibl id="bib1" idref="ref6" type="bt">1</bibl> <bibtext> Chen, B., Chang, Y.-H., Ouyang, F., & Zhou, W. (2018). Fostering student engagement in online discussion through social learning analytics. The Internet and Higher Education, 37 (2), 21 – 30. https://doi.org/10.1016/j.iheduc.2017.12.002</bibtext> </blist> <blist> <bibl id="bib2" idref="ref28" type="bt">2</bibl> <bibtext> Chen, L., & Wu, T. (2021). Short text sentiment analysis method combining topic model and self-attention mechanism. Foreign Electronic Measurement Technology, 40 (11), 18 – 23. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=GWCL202111004&DbName=DKFX2021</bibtext> </blist> <blist> <bibl id="bib3" idref="ref25" type="bt">3</bibl> <bibtext> Chen, X.-M., Zhang, C.-C., Wang, F., & Wang, X.-C. (2021). Feature dimension analysis of MOOC review response based on LDA topic model. Journal of Fuyang Normal University (Natural Science Edition), 38 (4), 73 – 81. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=FYSZ202104013&DbName=CJFQ2021</bibtext> </blist> <blist> <bibl id="bib4" idref="ref15" type="bt">4</bibl> <bibtext> Costley, J. (2015). The effects of three types of instructor posting on critical thinking and social presence: No posting, facilitating discourse, and direct instruction. International Journal of Learning, Teaching and Educational Research, 12 (2), 26 – 47.</bibtext> </blist> <blist> <bibl id="bib5" idref="ref48" type="bt">5</bibl> <bibtext> Farrokhnia, M., Banihashem, S. K., Noroozi, O., & Wals, A. (2023). A SWOT analysis of ChatGPT: Implications for educational practice and research. Innovations in Education and Teaching International, 1 – 15. https://doi.org/10.1080/14703297.2023.2195846</bibtext> </blist> <blist> <bibl id="bib6" idref="ref5" type="bt">6</bibl> <bibtext> Galikyan, I., Admiraal, W., & Kester, L. (2021). MOOC discussion forums: The interplay of the cognitive and the social. Computers & Education, 165 (6), 104133. https://doi.org/10.1016/J.COMPEDU.2021.104133</bibtext> </blist> <blist> <bibl id="bib7" idref="ref9" type="bt">7</bibl> <bibtext> Gao, F., Zhang, T., & Franklin, T. (2013). Designing asynchronous online discussion environments: Recent progress and possible future directions. British Journal of Educational Technology, 44 (3), 469 – 483. https://doi.org/10.1111/j.1467-8535.2012.01330.x</bibtext> </blist> <blist> <bibl id="bib8" idref="ref11" type="bt">8</bibl> <bibtext> Han, Z.-M., Huang, C.-Q., Yu, J.-H., & Tsai, C.-C. (2021). Identifying patterns of epistemic emotions with respect to interactions in massive online open courses using deep learning and social network analysis. Computers in Human Behavior, 122 (9), 106843. https://doi.org/10.1016/j.chb.2021.106843</bibtext> </blist> <blist> <bibl id="bib9" idref="ref14" type="bt">9</bibl> <bibtext> He, C., Ma, P., Zhou, L., & Wu, J. (2018). Is participating in MOOC forums important for students? A data-driven study from the perspective of the supernetwork. Journal of Data and Information Science, 3 (2), 62 – 77. https://doi.org/10.2478/jdis-2018-0009</bibtext> </blist> <blist> <bibtext> Jiang, S., & Wang, W. (2018). Interactive design of collaborative online courses: A case study of educational communication principles. China Distance Education, 5 (5), 35 – 44+7980. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=DDJY201805005&DbName=CJFQ2018</bibtext> </blist> <blist> <bibtext> Liang, C., & Li, R. (2020). Image perception analysis of Lijiang ancient city based on LDA and feature dimensions. Progress in Geography, 39 (4), 614 – 626. https://doi.org/10.18306/dlkxjz.2020.04.008</bibtext> </blist> <blist> <bibtext> Lin, F.-R., Lin, S.-C., & Huang, T.-P. (2018). Knowledge sharing and creation in a teachers'professional virtual community. Computers & Education, 50 (3), 742 – 756. https://doi.org/10.1016/j.compedu.2006.07.009</bibtext> </blist> <blist> <bibtext> Liu, Y. (2020). Social network analysis: A new perspective on supporting the role of collaboration. China Educational Technology and Equipment, 08, 5 – 9. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=ZJJB202008002&DbName=CJFQ2020</bibtext> </blist> <blist> <bibtext> Liu, Z., Zhang, N., Peng, X., Liu, S., Yang, Z., Peng, J., Su, Z., & Chen, J. (2022). Exploring the relationship between social interaction, cognitive processing and learning achievements in a MOOC discussion forum. Journal of Educational Computing Research, 60 (1), 132 – 169. https://doi.org/10.1177/07356331211027300</bibtext> </blist> <blist> <bibtext> Li, Y., Xiong, H., Yang, M., & Ye, J. (2021). Research on user group topic mining in virtual academic community based on social network analysis and LDA. Information Science, 39 (11), 110 – 116+132. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=QBKX202111015&DbName=DKFX2021</bibtext> </blist> <blist> <bibtext> Martin, F., Wang, C., & Sadaf, A. (2018). Student perception of helpfulness of facilitation strategies that enhance instructor presence, connectedness, engagement and learning in online courses. The Internet and Higher Education, 37 (2), 52 – 65. https://doi.org/10.1016/j.iheduc.2018.01.003</bibtext> </blist> <blist> <bibtext> Ouyang, F., & Scharber, C. (2017). The influences of an experienced instructor's discussion design and facilitation on an online learning community development: A social network analysis study. The Internet and Higher Education, 35 (4), 34 – 47. https://doi.org/10.1016/j.iheduc.2017.07.002</bibtext> </blist> <blist> <bibtext> Ozyurt, B., & Ali Akcayol, M. (2021). A new topic modeling based approach for aspect extraction in aspect based sentiment analysis: SS-LDA. Expert Systems with Applications, 168 (6), 114231. https://doi.org/10.1016/j.eswa.2020.114231</bibtext> </blist> <blist> <bibtext> Rosen-Zvi, M., Griffiths, T., Steyvers, M., & Smyth, P. (2021). The author-topic model for authors and documents. arXiv Preprint arXiv, 1207, 4169. https://doi.org/10.48550/arXiv.1207.4169</bibtext> </blist> <blist> <bibtext> Saqr, M., & Alamro, A. (2019). The role of social network analysis as a learning analytics tool in online problem based learning. BMC Medical Education, 19 (1), 1 – 11. https://doi.org/10.1186/s12909-019-1599-6</bibtext> </blist> <blist> <bibtext> Shi, Y. J. (2020). Research on user interest of Sina Weibo based on LDA Topic Model (Master dissertation). Shandong University. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=1020067933.nh&DbName=CMFD2020</bibtext> </blist> <blist> <bibtext> Wang, R. (2020). Design of Blended Learning Online Discussion Activity Based on the Perspective of Social Constructivism. Shenyang Normal University. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=1020741596.nh&DbName=CMFD2020</bibtext> </blist> <blist> <bibtext> Wang, Y. (2018). Online learning behavior analysis and application Research. Central China Normal University. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=1018233982.nh&DbName=CMFD2018</bibtext> </blist> <blist> <bibtext> Warren, A. N., & Paulus, T. M. (2020). Postgraduate students' accomplishment of epistemic positioning through personal experience in online discussion forums. Classroom Discourse, 11 (1), 22 – 40. https://doi.org/10.1080/19463014.2018.1562357</bibtext> </blist> <blist> <bibtext> Wong, G. K. W., Li, Y. K., & Xiaoyan, L. (2021). Visualizing the learning patterns of topic-based social interaction in online discussion forums: An exploratory study. Educational Technology Research & Development, 69 (5), 2813 – 2843. https://doi.org/10.1007/s11423-021-10040-5</bibtext> </blist> <blist> <bibtext> Wu, L., Li, J., Jin, C., & Gao, T. (2021). Study on the influence of different types of teacher feedback on learners' learning engagement in online discussion. China Education Informationization, 11, 14 – 18. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=JYXX202111004&DbName=CJFQ2021</bibtext> </blist> <blist> <bibtext> Wu, J.-Y., & Nian, M.-W. (2021). The dynamics of an online learning community in a hybrid statistics classroom over time: Implications for the question-oriented problem-solving course design with the social network analysisapproach. Computers & Education, 166, 104120. https://doi.org/10.1016/j.compedu.2020.104120</bibtext> </blist> <blist> <bibtext> Xie, K., DiTosto, G., Lu, L., & Cho, Y.-S. (2018). Detecting leadership in peer-moderated online collaborative learning through text mining and social network analysis. The Internet and Higher Education, 38, 9 – 17. https://doi.org/10.1016/j.iheduc.2018.04.002</bibtext> </blist> <blist> <bibtext> Xu, L., & Chen, S. (2017). Social cognitive network of learners' knowledge construction. Open Education Research, 23 (5), 102 – 112. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=JFJJ201705012&DbName=CJFQ2017</bibtext> </blist> <blist> <bibtext> Zhang, S., Gao, Q., Ma, X., Wei, Y.-T., & Yang, H. R. (2021). Modeling of learners' conversational behavior in private class forum. Research on Audio-Visual Education, 42 (11), 63 – 68+106. https://kns.cnki.net/kcms/detail/detail.aspx?FileName=DHJY202111011&DbName=CJFQ2021</bibtext> </blist> <blist> <bibtext> Zhang, J., & Han, Y. (2021). A survey on the learning interaction status of MOOC courses in Chinese universities. Open Education Research, 27 (5), 73 – 80. https://kns.cnki.net/kcms2/article/abstract?v=3uoqIhG8C44YLTlOAiTRKibYlV5Vjs7iy_Rpms2pqwbFRRUtoUImHRzW2qL9OZ4NCD-CATIopvBCmcFifB4pfEx8fT9TWxPz&uniplatform=NZKPT</bibtext> </blist> </ref> <aug> <p>By Wei Xu; Yuhan Chen and Leying Yang</p> <p>Reported by Author; Author; Author</p> <p></p> <p>Wei Xu is an associate professor at the College of Educational Science and Technology of Zhejiang University of Technology. Her current research interests include instruction system design, museum learning, and the deep integration of information technology and curriculum. She has published papers in the Journal of Distance Education, China Educational Technology, and Modern Educational Technology.</p> <p>Yuhan Chen is a postgraduate student at the Department of Education Information Technology of East China Normal University. Her current research interests include artificial intelligence technology application in education.</p> <p>Le-Ying Yang is a postgraduate student at the College of Educational Science and Technology of Zhejiang University of Technology. Her current research interests include instruction system design and the deep integration of information technology and curriculum.</p> </aug> <nolink nlid="nl1" bibid="bib26" firstref="ref1"></nolink> <nolink nlid="nl2" bibid="bib22" firstref="ref2"></nolink> <nolink nlid="nl3" bibid="bib13" firstref="ref3"></nolink> <nolink nlid="nl4" bibid="bib23" firstref="ref4"></nolink> <nolink nlid="nl5" bibid="bib31" firstref="ref8"></nolink> <nolink nlid="nl6" bibid="bib25" firstref="ref10"></nolink> <nolink nlid="nl7" bibid="bib24" firstref="ref12"></nolink> <nolink nlid="nl8" bibid="bib14" firstref="ref13"></nolink> <nolink nlid="nl9" bibid="bib27" firstref="ref17"></nolink> <nolink nlid="nl10" bibid="bib10" firstref="ref18"></nolink> <nolink nlid="nl11" bibid="bib17" firstref="ref19"></nolink> <nolink nlid="nl12" bibid="bib29" firstref="ref20"></nolink> <nolink nlid="nl13" bibid="bib28" firstref="ref21"></nolink> <nolink nlid="nl14" bibid="bib30" firstref="ref26"></nolink> <nolink nlid="nl15" bibid="bib21" firstref="ref27"></nolink> <nolink nlid="nl16" bibid="bib19" firstref="ref29"></nolink> <nolink nlid="nl17" bibid="bib18" firstref="ref30"></nolink> <nolink nlid="nl18" bibid="bib15" firstref="ref33"></nolink> <nolink nlid="nl19" bibid="bib12" firstref="ref34"></nolink> <nolink nlid="nl20" bibid="bib438" firstref="ref36"></nolink> <nolink nlid="nl21" bibid="bib178" firstref="ref39"></nolink> <nolink nlid="nl22" bibid="bib248" firstref="ref41"></nolink> <nolink nlid="nl23" bibid="bib11" firstref="ref42"></nolink> <nolink nlid="nl24" bibid="bib20" firstref="ref43"></nolink> <nolink nlid="nl25" bibid="bib16" firstref="ref46"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1458056
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: The Dynamics of Social Performance and Cognitive Depth between Students and Teacher in Online Discussion Forums with the SNA and LDA Approach
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Wei+Xu%22">Wei Xu</searchLink> (ORCID <externalLink term="https://orcid.org/0000-0002-9042-213X">0000-0002-9042-213X</externalLink>)<br /><searchLink fieldCode="AR" term="%22Yuhan+Chen%22">Yuhan Chen</searchLink><br /><searchLink fieldCode="AR" term="%22Leying+Yang%22">Leying Yang</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22Innovations+in+Education+and+Teaching+International%22"><i>Innovations in Education and Teaching International</i></searchLink>. 2025 62(1):135-151.
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Routledge. Available from: Taylor & Francis, Ltd. 530 Walnut Street Suite 850, Philadelphia, PA 19106. Tel: 800-354-1420; Tel: 215-625-8900; Fax: 215-207-0050; Web site: http://www.tandf.co.uk/journals
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 17
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2025
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– Name: Audience
  Label: Education Level
  Group: Audnce
  Data: <searchLink fieldCode="EL" term="%22Higher+Education%22">Higher Education</searchLink><br /><searchLink fieldCode="EL" term="%22Postsecondary+Education%22">Postsecondary Education</searchLink>
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Cognitive+Processes%22">Cognitive Processes</searchLink><br /><searchLink fieldCode="DE" term="%22Electronic+Learning%22">Electronic Learning</searchLink><br /><searchLink fieldCode="DE" term="%22Discussion+%28Teaching+Technique%29%22">Discussion (Teaching Technique)</searchLink><br /><searchLink fieldCode="DE" term="%22Computer+Mediated+Communication%22">Computer Mediated Communication</searchLink><br /><searchLink fieldCode="DE" term="%22Teacher+Student+Relationship%22">Teacher Student Relationship</searchLink><br /><searchLink fieldCode="DE" term="%22Student+Role%22">Student Role</searchLink><br /><searchLink fieldCode="DE" term="%22Teacher+Role%22">Teacher Role</searchLink><br /><searchLink fieldCode="DE" term="%22Differences%22">Differences</searchLink><br /><searchLink fieldCode="DE" term="%22Prompting%22">Prompting</searchLink><br /><searchLink fieldCode="DE" term="%22Student+Participation%22">Student Participation</searchLink><br /><searchLink fieldCode="DE" term="%22Academic+Achievement%22">Academic Achievement</searchLink><br /><searchLink fieldCode="DE" term="%22Change%22">Change</searchLink><br /><searchLink fieldCode="DE" term="%22College+Students%22">College Students</searchLink><br /><searchLink fieldCode="DE" term="%22College+Faculty%22">College Faculty</searchLink><br /><searchLink fieldCode="DE" term="%22Social+Networks%22">Social Networks</searchLink><br /><searchLink fieldCode="DE" term="%22Graphic+Arts%22">Graphic Arts</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1080/14703297.2023.2282155
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 1470-3297<br />1470-3300
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Online discussion forums are crucial educational tools that facilitate interaction between students and teachers. We created an online discussion forum, with the teacher serving as the moderator. 58 posts and 1,955 comments were collected. We analysed these data through Latent Dirichlet Allocation to determine word cooccurrence and extract key topics from the discussions. We examined how the topics and depth of the discussion changed over time and the roles that the students and teacher played in the forum by using time-series data. The results indicate that the teacher played a crucial role in the forum by initially serving as a moderator and then guiding or prompting students to discuss certain topics. Those who participated more actively achieved higher grades, and those that were passive had lower grades at the end Overall, the online discussion forum enhanced the course by helping students understand the core materials and topics in the course.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2025
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1458056
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1458056
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1080/14703297.2023.2282155
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 17
        StartPage: 135
    Subjects:
      – SubjectFull: Cognitive Processes
        Type: general
      – SubjectFull: Electronic Learning
        Type: general
      – SubjectFull: Discussion (Teaching Technique)
        Type: general
      – SubjectFull: Computer Mediated Communication
        Type: general
      – SubjectFull: Teacher Student Relationship
        Type: general
      – SubjectFull: Student Role
        Type: general
      – SubjectFull: Teacher Role
        Type: general
      – SubjectFull: Differences
        Type: general
      – SubjectFull: Prompting
        Type: general
      – SubjectFull: Student Participation
        Type: general
      – SubjectFull: Academic Achievement
        Type: general
      – SubjectFull: Change
        Type: general
      – SubjectFull: College Students
        Type: general
      – SubjectFull: College Faculty
        Type: general
      – SubjectFull: Social Networks
        Type: general
      – SubjectFull: Graphic Arts
        Type: general
    Titles:
      – TitleFull: The Dynamics of Social Performance and Cognitive Depth between Students and Teacher in Online Discussion Forums with the SNA and LDA Approach
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Wei Xu
      – PersonEntity:
          Name:
            NameFull: Yuhan Chen
      – PersonEntity:
          Name:
            NameFull: Leying Yang
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2025
          Identifiers:
            – Type: issn-print
              Value: 1470-3297
            – Type: issn-electronic
              Value: 1470-3300
          Numbering:
            – Type: volume
              Value: 62
            – Type: issue
              Value: 1
          Titles:
            – TitleFull: Innovations in Education and Teaching International
              Type: main
ResultId 1