Topic Sentiment Analysis in Online Learning Community from College Students

Kai Wang; Yu Zhang

Uneingeschränkter Zugang

Topic Sentiment Analysis in Online Learning Community from College Students

Kai Wang

und

Yu Zhang

| 20. Mai 2020

Journal of Data and Information Science

Band 5 (2020): Heft 2 (April 2020)

Über diesen Artikel

Vorheriger Artikel

Nächster Artikel

Zitieren

Article Category: Research Paper

Online veröffentlicht: 20. Mai 2020

Seitenbereich: 33 - 61

Eingereicht: 19. Sept. 2019

Akzeptiert: 23. März 2020

DOI: https://doi.org/10.2478/jdis-2020-0009

Schlüsselwörter
Online learning community, Topic detection, Sentiment analysis

© 2020 Kai Wang et al., published by Sciendo

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

Schematic diagram of topic generation based on LDA model.

The overall structure of the proposed methodology.

A screenshot of the tool of documents-topics for sentiment mining.

Precision (left) and recall (right) comparison based on various classifiers.

F-measure (left) and MAE (right) comparison based on various classifiers.

A proposed method algorithm for calculating sentiment scores matrix.

Input:	A topic formal context K=(U, T, I), where U={u₁,u₂,…,u_n} represents a set of topics belonging to a group of students, T={t₁,t₂,…,t_m}, n is the size of student set, m is the size of topics.
Output:	A sentiment scores matrix Sentiment_score(t_i), where irepresents sentiment score of each topic.
1.	for each topic t_i in T.
2.	Sentiment_score(t_i)=0.
3.	P(t_i, u_i)=0.
4.	Derive the positive and negative seed terms on the basis of domain experts.
5.	Compute sim_KL(t_i, u_j) // Compute the mutual information.
6.	Compute SD(t_i, t_seed) // Compute the sentiment comprehensive value.
7.	for each topic of student u_j in the topic formal context K.
8.	SD(t_i, u_j)=0.
9.	for each topic of t_i in the topic formal context K.
10.	SD(t_i, u_j)= SD(t_i, t_seed)+ SD(u_j, t_seed).
11.	end for.
12.	Sentiment_score(t_i)= Sentiment_score(t_i)+ SD(t_i, u_j).
13.	end for.
14.	end for.
15.	Return Sentiment_score(t_i).

The binary sentiment of the single-valued formal context.

	T₁	T₂	T₃	T₄	T₅	T₆	T₇	T₈	T₉	T₁₀	T₁₁	T₁₂	T₁₃	T₁₄	T₁₅	T₁₆	T₁₇	T₁₈	T₁₉	T₂₀
D₁	^*			^*			^*	^*		^*			^*				^*	^*
D₂			^*		^*	^*			^*	^*		^*		^*	^*					^*
D₃				^*			^*	^*			^*		^*		^*			^*	^*
D₄	^*				^*	^*		^*			^*	^*	^*		^*
D₅		^*			^*	^*			^*	^*		^*		^*			^*			^*
D₆		^*			^*	^*		^*			^*		^*				^*	^*
D₇			^*		^*		^*	^*			^*			^*		^*				^*
D₈	^*				^*		^*	^*			^*	^*	^*		^*			^*
D₉	^*				^*	^*		^*		^*			^*		^*					^*
D₁₀			^*		^*	^*		^*		^*		^*	^*	^*	^*			^*

A proposed method algorithm for topic-clustered concept lattice generation.

Input:	A set of topic and comment documentation D, where \| D \|=n, the number of potential topics m.
Output:	A topic-clustered concept lattice CL, a topics-terms probability matrix P and a documents-topics probability matrix R.
1.	for each d_i ∈ D.
2.	d_i → CWS_i. // Convert the document into a word segment.
3.	for each cws in CWS_i.
4.	W = W ∪ {cws}. // Obtain a collection of phrases that contains topic attribute.
5.	end for.
6.	end for.
7.	for each cws in CWS_i.
8.	CWS_i → tfidf_i. // Calculate the term frequency of attributes.
9.	$\overset{‘}{D} = [D : {tfidf}_{i}]$ \mathop D\limits^{'} = [D\,\,:\,{tfidf}_i] . // Obtain term frequency vector.
10.	end for.
11.	$(\overset{‘}{D}, W) \overset{LDA}{\to} (D, P, I)$ (\mathop D\limits^{'} ,W)\buildrel {LDA} \over \longrightarrow (D,P,I) . // Perform topic detection.
12.	$(\overset{‘}{D}, P) \to R$ (\mathop D\limits^{'} ,\,P) \to R . // Classify topic association matrix.
13.	Find the subset of topic attributes represented as t_j.
14.	for j=1 to 2^m.
15.	Compute the set of objects by applying the Glois connection.
16.	R → I′. // Convert topic association matrix to multi-valued formal context.
17.	I′ → I. // Convert multi-valued formal context to binary single-valued formal context.
18.	$(\overset{‘}{D}, R, I') \overset{FCA}{\to} CL (D, R, I')$ (\mathop D\limits^{'} ,R,I')\buildrel {FCA} \over \longrightarrow CL(D,R,I') . // Construct a hierarchical topic concept lattice.
19.	end for.
20.	Return {CL( $\overset{‘}{D}$ \mathop D\limits^{'} , R, I′), P, R}.
21.	Derive the topic-clustered sets.

Classification weights for adverb of degree.

Level(weights)	Included adverbs
adv₁(1.5)	excessively, completely, extensively, dreadfully, entirely, absulutely
adv₂(1.3)	fairly, pretty, rather, quite, very, much, greatly, by far, hightly, deeply
adv₃(1.1)	really, almost, nearly, bven, just, still
adv₄(1)	slightly, a little, a bit, trifle, somewhat

Multi-valued sentiment formal context based on topic association matrix.

	T₁	T₂	T₃	T₄
D₁	−3.427	2.874	4.315	−1.306
D₂	2.641	−0.597	−2.105	2.635
D₃	4.715	2.132	1.624	0
D₄	2.334	0	−1.748	4.316
D₅	−3.619	−1.857	3.624	−0.391
D₆	−2.107	2.167	2.419	2.361
D₇	0	−0.524	−0.267	2.638
D₈	2.369	1.629	2.364	0
D₉	1.024	−0.121	3.478	2.964
D₁₀	2.361	1.493	−0.328	−1.267

The implication rules and association rules.

Association rules	1<3>Learner Information provider<AVG NT2=[100%]=><3>Information searcher>AVG;
	2<4>Learner Psychological stress PT1=[75%]=><3>Information provider<AVG NT3;
	3<4>Learner NT2 =[75%]=><3>Interaction;
	4<4>Learner NT2 =[75%]=><3>Information sharer>AVG Information searcher>AVG;
	5<3>Learner Information sharer<AVG Psychological stress Cooperation PT1=[67%]=><2> Information provider<AVG NT3;
	6<3>Learner Information searcher>AVG Psychological stress PT1 NT3 =[67%]=><2> Postgraduate Information searcher<AVG Interaction;
	7<3>Learner Information provider<AVG PT1 PT4=[67%]=><2>Information searcher<AVG NT2;
	8<3> Learner Information provider<AVG Information searcher<AVG NT2=[67%]=><2> Information sharer>AVG Psychological stress Interaction;
	9<3>Learner NT2 PT4 =[67%]=><2>Postgraduate Interaction;
	10<3>Learner NT2 PT4 =[67%]=><2>Information sharer<AVG;
Implication rules	1<2>Learner Information sharer>AVGInteraction cooperation ==> Information searcher>AVG Psychological stress PT2;
	2<2>Learner Interaction sharer>AVG NT3==> Information searcher<AVG Psychological stress;
	3<2>Learner Information searcher>AVG Interaction cooperation ==> Information sharer>AVG Psychological stress PT4;

Recognition results of topic terms.

Topic	Term and its probability
T₁	Course selection/0.023, Learning objectives/0.021, Difficulty of knowledge/0.018, Teaching methods/0.017, Guidance methods/0.013
T₂	Credits/0.025, Content organization/0.023, Teaching methods/0.021, Learning support/0.021, Homework and assessment methods/0.020
T₃	Case presentation/0.032, Procedural evaluation/0.031, Knowledge expansion/0.029, Analysis of difficult points/0.027, Group discussion/0.027
T₄	Communication and feedback/0.033, Resource sharing/0.033, Information update/0.032, Response time/0.031, Information acceptance/0.030

Precision contrast between different methods based on SVM.

	S_t1	S_t2	S_t3	S_t4	S_t5	S_t6	S_t7
RA	49.32	37.51	40.67	42.52	43.77	41.26	45.33
CG	52.33	34.96	38.79	41.68	40.17	37.74	42.59
CoT	57.73	46.28	48.85	44.84	51.39	47.77	48.25
TextBlob	58.86	45.16	46.07	42.33	52.78	45.56	52.63
TSAOLC	61.34	50.23	54.95	49.83	53.95	62.98	54.36

MAE contrast between different methods based on SVM.

	S_t1	S_t2	S_t3	S_t4	S_t5	S_t6	S_t7
RA	98.42	92.46	90.87	88.38	89.07	91.45	95.63
CG	82.03	85.56	87.69	89.06	92.61	94.97	86.36
CoT	78.84	76.34	72.19	68.78	75.43	76.35	78.62
TextBlob	72.93	67.45	69.37	64.92	70.14	68.62	62.15
TSAOLC	58.99	54.56	57.32	55.25	57.20	59.15	53.13

Recall contrast between different methods based on SVM.

	S_t1	S_t2	S_t3	S_t4	S_t5	S_t6	S_t7
RA	44.45	42.06	47.64	44.37	45.98	41.63	48.21
CG	42.68	40.97	48.86	42.07	43.63	42.88	47.71
CoT	49.99	47.38	52.84	55.36	52.09	49.23	53.84
TextBlob	54.18	45.84	51.67	58.07	62.29	53.46	60.06
TSAOLC	56.49	58.03	62.27	59.96	65.59	58.76	62.34

F-measure contrast between different methods based on SVM.

	S_t1	S_t2	S_t3	S_t4	S_t5	S_t6	S_t7
RA	46.67	39.65	43.88	43.43	44.85	41.44	46.73
CG	47.01	37.73	43.25	41.87	41.83	40.15	45.00
CoT	53.58	46.82	50.77	49.55	51.74	48.49	50.89
TextBlob	56.42	45.50	48.71	48.97	57.14	49.19	56.10
TSAOLC	58.82	53.85	59.38	54.43	59.20	60.80	58.08

eISSN:: 2543-683X
Sprache:: Englisch

Zeitrahmen der Veröffentlichung:: 4 Hefte pro Jahr
Fachgebiete der Zeitschrift:: Informatik, Informationstechnik, Projektmanagement, Datanbanken und Data Mining

Zeitschrift RSS Feed

Topic Sentiment Analysis in Online Learning Community from College Students

Article Category: Research Paper

Online veröffentlicht: 20. Mai 2020

Seitenbereich: 33 - 61

Eingereicht: 19. Sept. 2019

Akzeptiert: 23. März 2020

DOI: https://doi.org/10.2478/jdis-2020-0009

Schlüsselwörter
Online learning community, Topic detection, Sentiment analysis

© 2020 Kai Wang et al., published by Sciendo

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

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

A proposed method algorithm for calculating sentiment scores matrix.

The binary sentiment of the single-valued formal context.

A proposed method algorithm for topic-clustered concept lattice generation.

Classification weights for adverb of degree.

Multi-valued sentiment formal context based on topic association matrix.

The implication rules and association rules.

Recognition results of topic terms.

Precision contrast between different methods based on SVM.

MAE contrast between different methods based on SVM.

Recall contrast between different methods based on SVM.

F-measure contrast between different methods based on SVM.

Topic Sentiment Analysis in Online Learning Community from College Students

Article Category: Research Paper

Online veröffentlicht: 20. Mai 2020

Seitenbereich: 33 - 61

Eingereicht: 19. Sept. 2019

Akzeptiert: 23. März 2020

DOI: https://doi.org/10.2478/jdis-2020-0009

SchlüsselwörterOnline learning community, Topic detection, Sentiment analysis

© 2020 Kai Wang et al., published by Sciendo

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

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

A proposed method algorithm for calculating sentiment scores matrix.

The binary sentiment of the single-valued formal context.

A proposed method algorithm for topic-clustered concept lattice generation.

Classification weights for adverb of degree.

Multi-valued sentiment formal context based on topic association matrix.

The implication rules and association rules.

Recognition results of topic terms.

Precision contrast between different methods based on SVM.

MAE contrast between different methods based on SVM.

Recall contrast between different methods based on SVM.

F-measure contrast between different methods based on SVM.

Schlüsselwörter
Online learning community, Topic detection, Sentiment analysis