Sentiment analysis tool


Sentiment analysis tool is the process of analyzing the emotional tone of the comments, concerning tourist activities, made by customers of SMEs. Thus, the tool tracks the reviews in social media to analyze satisfaction (using proxy such as emotions, attitudes, interaction). The team of WP2 did not collect data for the satisfaction index using a survey because the index is to be created through the reviews and customers´ posts on the web pages. Therefore, during the months from March to August, the team was dedicated to collecting reviewers and preparing the algorithm for sentiment analysis. The algorithm relies on Natural Language Processing (NLP) techniques to interpret subjective and unstructured data. In general, the text is labeled as positive, neutral, or negative, with different degrees of confidence. Recall that lexicon-based systems require a pre-compilated sentiment lexicon corpus, where each word has an assigned sentiment value, and these lexicons can be either manually or automatically generated. In our case, the synonymous keywords for satisfaction mentioned above are manually generated. A lexicon is roughly a structure containing words and possible information about them. Then the polarity of the text can be computed by a weighted count of such lexical items. Hence, considering that we focus on a lexicon-based approach, we will be using the pre-designed state-of-art classifiers TextBlob and VADER (Valence Aware Dictionary and Sentiment Reasoner) to support the implementation of the sentiment analysis module. Textblob is a Python library that processes text at a sentence level, that is, it takes the text as input and then splits it into sentences (see reference below). Based on a rule-based model, it returns a tuple with values called polarity (counting of positive and negative sentences) and subjectivity (amount of effective information in a sentence). We use VADER, which is a specific text sentiment analysis package. Like TextBlob, VADER is a lexicon and rule-based classifier, so it depends on the lexicons of sentiment-related words. We will also rely on data from the NLTK platform, which provides over 50 corpora and lexical resources in the field of NLP.

Primary Goals

The main aim of creating a sentiment analysis tool is to empower entrepreneurs and municipalities to improve their performance by means of tracking their current indicators in terms of the emotional states of the customers, fostering best practices, and exchanging processes and solutions. Thus, SMEs and municipalities can analyze the evolution of the emotional state of the customer (positive, negative, or neutral) through the text (comments) that they left in webpages. This tool is incorporated in together with other tool, such as service quality, self-diagnostic sustainability, and forecasting.

Key Features

The key feature of the sentiment analysis tool is to allow SMEs and Municipalities to visualize the positive, negative, or neutral sentiments (emotions) related to the company, including the relative positioning when the overall companies are considered. The visualization of the graphics gives a direct image about how positive, negative or neutral are the emotions of the customers and how it evolves. SMEs and municipalities can analyse those graphics and strategically discuss and take decisions about keep the service provided how it is or made improvements. This tool together with others incorporated in give a complementary diagnostic to support the planification of the strategy and take decisions by managers in charge of SMEs and Municipalities.

Who can use and benefit from this app?

The tool sentiment analysis receives the reviews made by customers of the SMEs. Therefore, the audience is the customers. The SMEs and municipalities involved in the resetting project will benefit from this tool. Later those SMEs and municipalities will decide if they permit other enterprises or not. The sentiment analysis is a tool incorporated in to support SMEs and municipalities to analyze the actual situation of the service provided and strategically decide what they should do in the future to improve the service to their customers.but not directly related to the registered RESETTING stakeholders.

How Can I use it?

For instance, if the result of the sentiment analysis is negative, this means that the analysis of the comments of the customers (analyzed through the Natural Language Processing (NLP) technique incorporated in the platform) is unfavorable, that is, the customers are overall unsatisfied with the service provided. Thus, managers should take action to change the service provided to satisfy the customers. aggregates diverse tools and they should operate together for a better diagnosis. In this vein, this tool should be combined with the service quality tool to better understand what exactly the aspects of the service provided that should mainly be improved.