Abstract:
In Sri Lanka, the tourism industry acts as one of the main roles in its economic
growth. However, as this industry was heavily affected by the Covid-19
pandemic, this dominant income source of the country has become unstable.
Hence, if we could identify the issues associated with this, we can be ready for
the safety of the industry before any other pandemic in the future. Therefore,
studying people’s ideas becomes very important and Twitter is one of the best
places for collecting their ideas. This study proposes a model for the analysis
of the tourism industry-related tweets during Covid-19 in Sri Lanka into four
categories; positive, negative, advertisement, and neutral. Here, 18980 tweets
were collected from the period between years 2020 to 2022. After
preprocessing, 6257 tweets were selected and used for extracting the feature
vectors. Three different machine learning algorithms; Support Vector
Machine, Long Short-Term Memory, and Artificial Neural Networks (ANN)
were used to create a forecast paradigm and these were compared to find the
best model. Among the four percentages of training data (62%, 67%, 72%,
and 77%) 67% of training and 33% of the testing data set was selected.
According to the results, ANN’s 91% highest accuracy was achieved. Also,
the study considered precision, recall, f-measure, and error values for the
evaluation. According to the final result, the ANN algorithm showed the best
results for sentiment analysis for tracking tourism industry movement in Sri
Lanka. In this period, industry declined and negative sentiments help to
identify the main issues.
