Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.

v1

SATLabel: A Framework for Sentiment and

Aspect Terms Based Automatic Topic Labeling

Khandaker Tayef Shahriar1,* , Mohammad Ali Moni2 , Mohammed Moshiul

Hoque1 , Muhammad Nazrul Islam3 , Iqbal H. Sarker1,*
1
Department of Computer Science and Engineering, Chittagong University of
Engineering & Technology, Chittagong-4349, Bangladesh.
2
Artificial Intelligence & Digital Health Data Science, School of Health and
Rehabilitation Sciences, Faculty of Health and Behavioural Sciences, The University
of Queensland St Lucia, QLD 4072, Australia.
3
Department of Computer Science and Engineering, Military Institute of Science and
Technology, Dhaka-1216, Bangladesh.
∗
Correspondence: [email protected], [email protected]

Abstract. In this paper, we present a framework that automatically

labels Latent Dirichlet Allocation (LDA) generated topics using senti-
ment and aspect terms from COVID-19 tweets to help the end-users by
minimizing the cognitive overhead of identifying key topics labels. Social
media platforms especially Twitter are considered as one of the most
influential sources of information for providing public opinion related to
a critical situation like the COVID-19 pandemic. LDA is a popular topic
modelling algorithm that extracts hidden themes of documents without
assigning a specific label. Thus automatic labelling of LDA-generated
topics from COVID-19 tweets is a great challenge instead of following
the manual labelling approach to get an overview of wider public opin-
ion. To overcome this problem, in this paper, we propose a framework
named SATLabel that effectively identifies significant topic labels using
top unigrams features of sentiment terms and aspect terms clusters from
LDA generated topics of COVID-19 related tweets to uncover various is-
sues related to the COVID-19 pandemic. The experimental results show
that our methodology is more effective, simpler, and traces better topic
labels compare to the manual topic labelling approach.

Keywords: Data-driven Framework · LDA · Sentiment Terms · Aspect

Terms · Unigrams · Soft Cosine Similarity · Topic · Automatic labeling

1 Introduction

Twitter nowadays is considered as one of the most important social media plat-
forms to explain the characteristics and predict the status of the pandemic [9].
In Wuhan, at the end of 2019, a novel coronavirus disease that causes COVID-
19 was reported by the World Health Organisation (WHO). The declaration
of COVID-19 as an international concern of public health emergency by WHO

© 2022 by the author(s). Distributed under a Creative Commons CC BY license.

Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

2 K. T. Shahriar et al.

was reported on January 30, 2020 [1]. During the pandemic, the use of Twit-
ter increases immensely and plays a critical role by reflecting real-time public
panic and providing rich information to raise public awareness through posts
and comments. However, text mining and analysis of data from social media
platforms such as Twitter have become a burning issue to extract necessary in-
formation. Moreover, it is a great challenge to extract meaningful topic labels
by machines instead of following diverse human interpretations of the manual
labelling approach [?]. Hence, in this paper, we propose SATLabel, a framework
that effectively identifies key topic labels of tweets automatically from the huge
volume of the Twitter dataset to reduce the human effort of cumbersome topic
labelling tasks.
A large number of labelled datasets is required for traditional supervised
methods. Obtaining such a labelled dataset for topic labelling purposes is very
difficult and expensive. In this paper, we use LDA [3], which is an unsupervised
probabilistic algorithm for text documents. Thus SATLabel does not need any
labelled dataset for topics. A set of topics available in the documents is discov-
ered by LDA. Sentiment terms express emotions from tweets and Aspect terms
describe features of an entity [19]. We create sentiment terms cluster and aspect
terms cluster for each LDA generated topic. However, Unigram is a probabilis-
tic language model that is extensively used in natural language processing tasks
and text mining to exhibit the context of texts. SATLabel uses the top Unigrams
features from sentiment terms cluster and aspect terms cluster respectively and
create attribute tags concatenating the two top Unigrams features (first senti-
ment term and then aspect term). We select the attribute tag which has the
highest soft cosine similarity value with respect to the tweets of the same topic
to assign a meaningful label for that LDA-generated topic. Our experimental re-
sults show that the label generated by SATLabel has a high soft cosine similarity
value with the tweets of the same topic than the manual labelling approach. The
main contributions of this paper can be summarized as follows:

– We effectively utilize sentiment terms and aspect terms of tweets to produce

significant topic labels.
– We propose a new framework named SATLabel that is useful to extract topics
from COVID-19 tweets and labels them automatically instead of following
the manual method.
– SATLabel effectively reduces the human effort for difficult topic labelling
tasks of tweets.
– We have shown the effectiveness of SATLabel comparing with the manual
labelling approach by conducting a range of experiments.

The organization of the rest of the paper is as follows. Related works are reviewed
in section 2. In section 3, we present the methodology of the proposed framework.
In section 4, we assess the evaluation results of our framework by conducting
experiments on the Twitter dataset. Next, we present the discussion, and finally,
we conclude this paper and highlight the direction of future work.
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

Automatic Topic Labeling 3

2 Related Work

COVID-19 tweets can be helpful for identifying meaningful topic labels to high-
light user conversation and understand ideas of people’s needs and interests.
Many researchers used the LDA algorithm to extract hidden themes of docu-
ments. Patil et. al. [12] proposed a paper using the frequency-based technique
to extract topics from people’s reviews without mentioning the proper labelling
techniques for describing the topics. Hingmire et. al. [7] proposed a paper to
construct LDA based topic model but the expert association is required to as-
sign the topic to the class labels. Hourani et. al. [8] proposed a paper to classify
articles according to their topics for which labelled dataset is required. Asmussen
et. al. [2] proposed a topic modelling method for researchers but topic labelling
depends on the researcher’s view without having any automatic method. Wang
et. al. [18] proposed a paper that minimizes the problem of data sparsity with-
out labelling key topics specifically. Zhu et. al. [20] presented the change of the
number of texts on topics with respect to time by following the manual topic
labelling approach. Satu et. al. [15] proposed a framework that extracts topics
from the best cluster of sentiment classification having a manual explanation
of topic labels tends to misinterpretation. Kee et. al. [10] used LDA to extract
higher-order arbitrary topics but only 61.3% clear collective themes were evalu-
ated. Maier et. al. [11] presented accessibility and applicability of communication
researchers using LDA based topic labelling approach which depends manually
on broader context knowledge. In our previous work, we only considered the top
unigram feature of aspect terms cluster to identify the key topics with labels
by implementing LDA [17]. Elgesem et. al. [5] presented an analysis about the
discussion of the Snowden affair using a manual topic labelling approach. Guo
et. al. [6] compared dictionary-based analysis and LDA analysis using a manual
topic labelling approach.
The summary of the above works describes that most of the works considered
a manual topic labelling approach to categorize documents and get an overview
which is expensive, time-consuming, and requires cumbersome human interpre-
tations. Hence, an automatic and effective topic labelling approach would be
helpful to reduce human effort and save time. Thus in this paper, we consider
the development of a framework named SATLabel to generate significant topic
labels automatically to highlight users’ conversations on Twitter.

3 Methodology

In this section, we present SATLabel that is a framework to label LDA gener-

ated topics automatically as shown in Fig. 1. For analyzing and mining textual
data like tweets, text preprocessing is one of the most essential steps to advance
in further processing steps. The working principle and overall steps to gener-
ate automatic topic labels from the Twitter dataset are shown in Algorithm 1.
After preprocessing of highly unstructured and non-grammatical tweets, several
processing steps are followed to produce the expected output.
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

4 K. T. Shahriar et al.

Fig. 1. SATLabel: Proposed framework for automatic topic labeling

Algorithm 1: Automatic Topic Labeling

Input: T: number of Tweets in dataset
Output: Topic Label (TLabel ).
1 for each t ∈ {1,2,...,T} do
2 Tp ← P reprocess(t);
3 for each tp ∈ {1,2,...,Tp } do
// Corpus Development
4 C ← Create Corpus(tp );
// Sentiment Terms and Aspect Terms Extraction
5 STp ← Sentiment T erms(tp );
6 ATp ← Aspect T erms(tp );
// Topic Discovery
7 K ∼ M allet(LDA(Doc2bow(C)));
8 for each k ∈ {1,2,...,K} do
9 for each tp ∈ {1,2,...,Tp } do
10 kdominant,tp ∼ dominant topic(tp , k);
// Create Clusters from Topic
11 CS ∼ Cluster(STp → kdominant,tp );
12 CA ∼ Cluster(ATp → kdominant,tp );
13 for each k ∈ {1,2,...,K} do
14 US ∼ max count(T op U nigrams(CS → k));
15 UA ∼ max count(T op U nigrams(CA → k));
16 TLabel ← max sof t cosine similarity(US + UA , k)

3.1 Sentiment and Aspect Terms Extraction

Sentiment terms carry the tone or opinion of the text. Usually, adjectives and
verbs of sentences are considered as sentiment terms that indicate expressed
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

Automatic Topic Labeling 5

opinion of the text. Noun and noun phrases are considered as aspects terms
of text. Objects of verbs are often regarded as aspect terms that describe the
features of an entity, product, or event [19]. We follow precise parts of speech
tagging which is an efficient approach to extract sentiment terms and aspect
terms from texts. Examples of sentiment terms and aspect terms of sample
tweets are shown in Table 1.

Table 1. Example of Sentiment Terms and Aspect Terms

Sentiment Aspect
Sample Tweet
Terms Terms
Please read the thread. read thread
To enjoy and relax for your dinner enjoy, relax,
dinner, place
it is a great place. great
Links with info on communicating with communicate, links, info,
children regarding COVID-19. covid children
The retail store owners right now retail, right owners, store

3.2 Topic Identification Using LDA

LDA is a popular topic modeling algorithm to discover hidden topics available
in the corpus from unlabelled dataset [14]. But the challenge is how to assign
significant labels to LDA-generated topics. The steps for topic discovery that we
follow in SATLabel framework are discussed below:
1) Creating Dictionary and Corpus: A systematic way of creating a number of
lexicons of a language is supported by a dictionary and a corpus generally
refers to an arbitrary sample of that language. A document corpus is built
with words or phrases. In Natural Language Processing (NLP) paradigm,
the corpus of a language plays a vital role in developing a knowledge-based
system and mining texts. In the proposed framework, we create a dictionary
and develop a corpus from the preprocessed text.
2) Creating a BoW Corpus: Corpus contains the word id and its frequency in
every document. Documents are converted into Bag of Words (BoW) format
by applying Doc2bow embedding. Each word is assumed as a normalized
and tokenized string.
3) Topic Discovery: BoW corpus is transferred to the mallet wrapper of LDA.
The presence of a set of topics in the corpus is discovered by LDA. Mallet
wrapper of LDA runs faster and provides precise division of topics using
Gibbs Sampling technique [4]. LDA generates the most prominent words in
a topic. Thus by using the word probabilities one can manually find dom-
inant themes in the documents. To overcome the complex manual labeling
approach our framework SATLabel generates automatic topic labels using
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

6 K. T. Shahriar et al.

sentiment terms and aspect terms of documents without any human in-
terpretation. Based on the topic coherence score, we choose a model that
discovers 20 optimal number of topics itself. Then we enumerate the domi-
nant topic for each tweet to understand the distribution of topics across the
tweets in the dataset.

3.3 Output Generation

The steps to generate significant topic labels automatically as output from the
topics extracted by LDA are discussed below:

1) Generation of Sentiment Terms and Aspect Terms Cluster: We create clus-

ters of sentiment terms and aspect terms independently from tweets cor-
responding to each LDA-generated topic. Thus, we get 20 sentiment terms
cluster and 20 aspect terms cluster from the discovered topics by LDA.
2) Labeling Topic using Top Unigrams: A Unigram is a one-word sequence of
n-gram. The use of unigrams can be observed in NLP, cryptography, and
mathematical analysis. Soft cosine similarity considers the similarity of fea-
tures in vector space model [16]. We extract the top 20 unigrams from senti-
ment terms cluster and aspect terms cluster respectively for each topic. Then
we concatenate all the possible combinations of top unigrams of sentiment
terms and top unigrams of aspect terms of topic. We select a combination
that has the highest soft cosine similarity value with respect to the tweets
of that topic to assign with a significant topic label. We use a sentiment
and aspect term tag to label each topic because that feature tag presents an
attribute to describe that topic of tweets.

In the section of the methodology of this paper, we present a framework called

SATLabel to detect key topic labels from the tweets automatically as shown in
Table 2. We compare the quality of topic labels generated by SATLabel with the
manually assigned topic labels in the experiment section. To categorize a tweet
with a specific topic label from test data, we search the topic number that has
a greater impact of percentage on that tweet.

4 Experiments

4.1 Dataset

We collect the Twitter dataset from the website at https://www.kaggle.com/

datatattle/covid-19-nlp-text-classification. There are two csv files in
the dataset. One is Corona NLP train.csv and another is Corona NLP test.csv.
Tweets available in the dataset are highly unstructured and non-grammatical
in syntax. There are 41,157 and 3,798 COVID-19 related tweets are available in
Corona NLP train.csv and Corona NLP test.csv files respectively. We apply a
series of preprocessing functions to get the normalized form of noisy tweets for
further processing.
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

Automatic Topic Labeling 7

4.2 Data Preprocessing

Handling ill-formatted, noisy and unstructured twitter data is one of the most
important tasks for us. We preprocess the Twitter dataset to get the normalized
form using the functions of transforming words into lowercase, replacing hy-
perlinks, mentions, and hashtags with empty string, dealing with contractions,
replacing punctuation with space, striping space from words, removing words less
than two characters, removing stop words, handling Unicode and non-English
words.

4.3 Finding the optimal number of topics for LDA

We create a function to return several LDA models with multiple values of a
number of topics (k) to find the optimal number of topics. The interpretable
topics can be found by selecting a ’k’ that identifies the end of a quick rise of
topic coherence score. Sometimes we get more granular sub-topics by choosing
a higher value of topic coherence score. We pick the model giving the highest

Fig. 2. Selection of the optimal number of LDA topics

coherence value before flattening out considering better sense while the coherence
score seems to keep growing as shown in Fig. 2. For the next steps, we choose
the model having 20 topics itself.

4.4 Selection of Top Unigrams Features from Clusters

We create sentiment terms cluster and aspect terms cluster of tweets for each
topic. We find the top counted 20 unigrams from each cluster. Fig. 3 and Fig. 4
show the top 20 unigrams from sentiment and aspect terms clusters of topic
no. 12 respectively. Then we detect the topic label depending on the highest
soft cosine similarity value of sentiment and aspect term tag with respect to the
tweets of that topic.
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

8 K. T. Shahriar et al.

Fig. 3. Top 20 unigrams from sentiment terms cluster of topic no. 12

Fig. 4. Top 20 unigrams from aspect terms cluster of topic no. 12

Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

Automatic Topic Labeling 9

Table 2. Example of Topics Detected on Tweets

Detected Topic Label

Sample Tweet Topic No.
(SATLabel)
Due to the COVID-19 virus and the
global health pandemic, we will be closed 17 Shut Location
at our retail location until further notice.
Dubai Becomes Cheaper To Live In. 9 Drop Cost
covid-19 is already affecting the online
16 Online Shopping
shopping, ok somebody slap meee plsss ?
You guys still can buy food during
0 Hoard Food
lockdown then why need to do panic buying?
I’m going to try patenting my world-famous
14 Learn Scam
vegetable phall as a killer of covid-19.
It’s Not Covid 19. It’s due fall in global oil
15 Drop Barrel
prices Oil cost 30 barrel...
Here’s a buying guide our community
set up for the neighborhood supermarket. 12 Covid Product
Feel free to use it as a template.
The Consumer Financial Protection Bureau
today announced that it is postponing some 3 Learn Insight
data collection from the financial industry.
Food demand in poorer countries is more
6 Covid Food
linked to income...

4.5 Qualitative Evaluation of Topic Labels

An expert annotator assigns the topic labels manually using the word probabil-
ities in LDA-generated topics to a randomly selected set of tweets. In Table 2,
we present a portion of set of tweets assigned by the SATLabel generated topic
labels. Table 2 shows that SATLabel generated topic labels are well-aligned and
closely coherent with the descriptions of tweets. We can extract useful informa-
tion related to a topic, simply by categorizing the tweets using the key label
generated by SATLabel of that topic.

4.6 Effectiveness Analysis

In this experiment section, we calculate the Soft Cosine Similarity (SCS) values
of detected topic labels by SATLabel and manual approach for LDA-generated
20 topics. SCS is used to detect the semantic text similarities between two doc-
uments. A high SCS value provides a high similarity index and similarity is
smaller for unrelated documents. We train the word2vec embedding model to
use SCS. We show the comparison of SATLabel and manual labeling approach
for all LDA-generated topics in terms of SCS value in Fig. 5. We get SCS values
generated by proposed SATLabel for topic no. 4, 8, 10, 14 are 0.77, 0.61, 0.53,
0.64 respectively while manual approach generates 0.09, 0.06, 0.02, 0.07 SCS
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

10 K. T. Shahriar et al.

Fig. 5. Comparison of SATLabel with manual approach

scores for those topics which are very low. Diverse human interpretation of top-
ics is the possible reason for the high difference of SCS scores between proposed
SATLabel and manual approach. For topics no. 3, 6, 9, 12, 16, 18 we get the
same scs values for SATLabel and manual approach because of identical topic
labels generated by both approaches. From Fig. 5, we can observe that the topic
labels generated by the proposed framework SATLabel produce high SCS values
for a maximum number of topics compared with the manual labeling approach.
Hence, our proposed framework is more effective and traces better topic labels
from unlabelled datasets to reduce the cumbersome task of the human manual
labeling approach.

5 Discussion

Automatic labeling of LDA-generated topics of the tweets of social media plat-

forms like Twitter is helpful to understand people’s ideas and feelings by going
through meaningful insights rather than following traditional strategies like the
manual labeling approach. In this paper, we use LDA, a popular probabilistic
topic modeling algorithm to extract hidden topics from tweets. We then effec-
tively use sentiment terms and aspect terms of tweets to create clusters. After
that, we select top unigrams from the clusters to produce significant topic la-
bels using the maximum soft cosine similarity values. Our proposed framework
SATLabel helps to produce semantically similar topic labels of tweets to highlight
the user’s conversations and notice several COVID-19 related issues.
Overall, SATLabel is a data-driven framework for topic labeling purposes
for mining texts to provide helpful information from the dataset of Twitter
related to COVID-19. We firmly believe that SATLabel can be effectively used
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

Automatic Topic Labeling 11

in other domains of applications like agriculture, healthcare, education, business,

cyber-security, etc, and also can be used to generate target class from unlabeled
datasets to train deep learning models [13]. These types of contributions allow
the researchers and experts in relevant departments to take necessary actions
in critical situations like the COVID-19 pandemic by efficiently utilizing social
media platforms.

6 Conclustion and Future Work

In this paper, we propose a new framework named SATLabel that effectively and
automatically identifies key topic labels from COVID-19 tweets. Our framework
saves time and reduces the human effort to minimize the overhead of difficult
topic labeling tasks from the huge volumes of data to get an overview of broader
public opinions on social media platforms like Twitter. We believe that SATLabel
will help the reformists to discover various COVID-19 related issues by analyzing
automatically extracted topic labels.
In the future, we want to increase our scope of experiments by integrating
the proposed framework with sentiment classification tasks using hybridization
of deep learning methods. We will also implement our proposed framework to
other social media platforms on different events to generate significant topic
labels to handle the overload of ever-increasing data volume.

References
1. Adhikari, S.P., Meng, S., Wu, Y.J., Mao, Y.P., Ye, R.X., Wang, Q.Z., Sun, C.,
Sylvia, S., Rozelle, S., Raat, H., et al.: Epidemiology, causes, clinical manifestation
and diagnosis, prevention and control of coronavirus disease (covid-19) during the
early outbreak period: a scoping review. Infectious diseases of poverty 9(1), 1–12
(2020)
2. Asmussen, C.B., Møller, C.: Smart literature review: a practical topic modelling
approach to exploratory literature review. Journal of Big Data 6(1), 1–18 (2019)
3. Blei, D.M., Ng, A.Y., Jordan, M.I.: Latent dirichlet allocation. the Journal of
machine Learning research 3, 993–1022 (2003)
4. Boussaadi, S., Aliane, H., Abdeldjalil, P.O.: The researchers profile with topic
modeling. In: 2020 IEEE 2nd International Conference on Electronics, Control,
Optimization and Computer Science (ICECOCS). pp. 1–6. IEEE (2020)
5. Elgesem, D., Feinerer, I., Steskal, L.: Bloggers’ responses to the snowden affair:
Combining automated and manual methods in the analysis of news blogging. Com-
puter Supported Cooperative Work (CSCW) 25(2-3), 167–191 (2016)
6. Guo, L., Vargo, C.J., Pan, Z., Ding, W., Ishwar, P.: Big social data analytics in
journalism and mass communication: Comparing dictionary-based text analysis
and unsupervised topic modeling. Journalism & Mass Communication Quarterly
93(2), 332–359 (2016)
7. Hingmire, S., Chougule, S., Palshikar, G.K., Chakraborti, S.: Document classifi-
cation by topic labeling. In: Proceedings of the 36th international ACM SIGIR
conference on Research and development in information retrieval. pp. 877–880
(2013)
Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 April 2022 doi:10.20944/preprints202204.0026.v1

12 K. T. Shahriar et al.

8. Hourani, A.S.: Arabic topic labeling using naı̈ve bayes (nb). In: 2021 12th In-
ternational Conference on Information and Communication Systems (ICICS). pp.
478–479. IEEE (2021)
9. Jahanbin, K., Rahmanian, V., et al.: Using twitter and web news mining to predict
covid-19 outbreak. Asian Pacific Journal of Tropical Medicine 13(8), 378 (2020)
10. Kee, Y.H., Li, C., Kong, L.C., Tang, C.J., Chuang, K.L.: Scoping review of mind-
fulness research: A topic modelling approach. Mindfulness 10(8), 1474–1488 (2019)
11. Maier, D., Waldherr, A., Miltner, P., Wiedemann, G., Niekler, A., Keinert, A.,
Pfetsch, B., Heyer, G., Reber, U., Häussler, T., et al.: Applying lda topic modeling
in communication research: Toward a valid and reliable methodology. Communi-
cation Methods and Measures 12(2-3), 93–118 (2018)
12. Patil, P.P., Phansalkar, S., Kryssanov, V.V.: Topic modelling for aspect-level sen-
timent analysis. In: Proceedings of the 2nd International Conference on Data En-
gineering and Communication Technology. pp. 221–229. Springer (2019)
13. Sarker, I.H.: Deep learning: A comprehensive overview on techniques, taxonomy,
applications and research directions. SN Computer Science 2(6), 1–20 (2021)
14. Sarker, I.H.: Machine learning: Algorithms, real-world applications and research
directions. SN Computer Science 2(3), 1–21 (2021)
15. Satu, M.S., Khan, M.I., Mahmud, M., Uddin, S., Summers, M.A., Quinn, J.M.,
Moni, M.A.: Tclustvid: a novel machine learning classification model to investigate
topics and sentiment in covid-19 tweets. Knowledge-Based Systems 226, 107126
(2021)
16. Sidorov, G., Gelbukh, A., Gómez-Adorno, H., Pinto, D.: Soft similarity and soft co-
sine measure: Similarity of features in vector space model. Computación y Sistemas
18(3), 491–504 (2014)
17. Tayef Shahriar, K., Sarker, I.H., Nazrul Islam, M., Moni, M.A.: A dynamic topic
identification and labeling approach of covid-19 tweets. In: International Confer-
ence on Big Data, IoT and Machine Learning (BIM 2021). Taylor and Francis
(2021)
18. Wang, B., Liakata, M., Zubiaga, A., Procter, R.: A hierarchical topic modelling
approach for tweet clustering. In: International Conference on Social Informatics.
pp. 378–390. Springer (2017)
19. Wang, W., Pan, S.J., Dahlmeier, D., Xiao, X.: Coupled multi-layer attentions for
co-extraction of aspect and opinion terms. In: Proceedings of the AAAI Conference
on Artificial Intelligence. vol. 31 (2017)
20. Zhu, B., Zheng, X., Liu, H., Li, J., Wang, P.: Analysis of spatiotemporal character-
istics of big data on social media sentiment with covid-19 epidemic topics. Chaos,
Solitons & Fractals 140, 110123 (2020)