Sentiment Analysis in Online Food Delivery: A Greek Case Study
- Abstract
Nikolaos Fragkos 1, Anastasios Liapakis 1, Constantina Costopoulou 1 and Maria Ntaliani 1
1
    Agricultural University of Athens, Iera Odos 75s, Athens, Greece


                 Summary
                 In recent years, a rapid and exponential increase in the volume of data is witnessed across
                 various sectors and platforms, such as the retail industry and social networks respectively. Most
                 of these data regard user-generated text, often including preferences, reviews or critics, and
                 above all underlying sentiments. However, this data should be elaborated to contribute to
                 business intelligence. Towards this end, sentiment analysis can help to obtain knowledge from
                 this hidden “big data treasure” for businesses.
                 Sentiment Analysis (or Opinion Mining) refers to the computational approach of identifying
                 the opinion or sentiment towards an entity and determining its polarity (positive, negative,
                 neutral). It uses a combination of machine learning and lexical resources as lexicons to achieve
                 the determination of the polarity. It offers the capability of creating a consumer profile by
                 monitoring personal preferences and beliefs, as well as the way of perceiving different trends,
                 to prompt the companies to build a new or modify an already existing strategy based on the
                 intelligence gathered by the analysis. Gathering such information results in better
                 understanding and serving customers, more efficient marketing, improving the quality of
                 products and services and augmenting profit margins. So far, the approach of sentiment
                 analysis has been implemented not only in business sectors, such as health care, tourism, and
                 banking, but also by governments in order to monitor social media and discover citizens’
                 concerns and sentiments.
                 In this context, this paper studies the potential of sentiment analysis for assessing consumer
                 behavior in online food delivery. A literature review is conducted on the implementation of
                 sentiment analysis approach in the food and beverage industry, including articles published
                 between 2011 and 2021 in the scientific databases Scopus, Web of Science and Wiley Online
                 Library. Then an overview of the top-ranked sentiment analysis tools is given including both
                 platforms and libraries for programming languages. These tools can help identify consumers’
                 needs. Moreover, sentiment analysis was conducted in 300 online consumers’ reviews for
                 online ordering, which were mined from one of the leading online food delivery platforms
                 operating in Greece. The assessment regards a sample of reviews and critics in Greek. More
                 specifically according to the results the tools used for the analysis were not very precise
                 compared to the analysis of a human agent, a fact that creates the need of specialized lexical
                 resources for each domain. Furthermore, it was observed that consumers are concerned mainly
                 about the quality of their order, the delivery speed and the service of the restaurant. These
                 results provide insights to food online retailers on the usefulness of sentiment analysis for
                 improving the overall quality of their services, customer relationship management, as well as
                 their business strategy.

                 Keywords 1
                 Sentiment Analysis, Food and Beverage Industry, Online Food Delivery, Greece




Proceedings of HAICTA 2022, September 22–25, 2022, Athens, Greece
EMAIL: fragosnick@gmail.com (A. 1); liapakisanastasios@aua.gr (A. 2); tina@aua.gr (A. 3); ntaliani@aua.gr (Α. 4)
ORCID: 0000-0003-0151-5586 (A. 3); 0000-0001-7643-5009 (A. 4)
              ©️ 2022 Copyright for this paper by its authors.
              Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
              CEUR Workshop Proceedings (CEUR-WS.org)




                                                                                    17