Abstract
The growth in World Wide Web (WWW) has made it difficult for any user to access the information they require, quickly and simply. With such demands and artificial intelligence technology advances, many companies have recently launched chatbot services. A chatbot can be considered a question–answer system in which experts provide knowledge on users’ behest. It often acts as a personal assistant also. Collaboration between artificial and human intelligence is necessary to increase and improve the ease of user interactions with systems. If it does, the chatbot may progress to a real virtual human over time. This study has tried to investigate the issue of and consumers’ reactions to chatbot services through online news and social media data. Through text mining, this study analyzes the positive and negative expressions about chatbot services. Ultimately, this study will be useful for future research on consumer satisfaction with chatbots. This paper can also possibly analyze and seek implications of the online news and social media perspective.
You have full access to this open access chapter, Download conference paper PDF
Similar content being viewed by others
Keywords
1 Introduction
A chatbot is a software designed to simulate an intelligent conversation with a human partner [1]. With improvement in data mining and machine learning techniques, better decision-making capabilities, availability of corpora, and robust linguistic annotation/processing tool standards such as XML and its applications, chatbots have become more practical and gained many commercial applications. Many companies are integrating chatbots into their websites to provide a better user experience. In South Korea, many companies have recently launched chatbot services in various industries such as finance, airlines, manufacturing, and food industries. Consumers have shown interest in launch of chatbot services, and the mention of chatbots has also increased in various articles and on social media.
This study investigated how people feel about chatbot services, inquired the buzz volume of chatbot services on online news and social media in South Korea, and compared positive and negative keywords revealed in each media. Analyzing the chatbot services online text data enabled the investigation of the emerging attributes and factors that must be considered in its future service development.
2 Method
This study classified as positive or negative the top 10 keywords collected from online news and social media data through text mining, also known as text data mining [2] or knowledge discovery from textual databases [3]. It generally refers to the process of extracting interesting and non-trivial patterns or knowledge from unstructured text documents and can be viewed as an extension of data mining or knowledge discovery from (structured) databases [4].
Text data were collected from three online news media and four social media using Trendup 3.0, which is a big online data collection engine. From January 1, 2016 to December 31, 2017, 1,438 news articles and 15,687 social media mentions were collected; R 3.4.1 was used to analyze.
3 Results
The results were as follows:
First, we extracted the most frequently mentioned industries regarding chatbot services in South Korea. The result shows that financial businesses were mentioned most, followed by communication businesses, as shown in Fig. 1.
Second, the number of online texts has been gradually increasing since 2016, as shown in Fig. 2. In particular, there was a sharp increase during the third quarter of 2016 when there was the major issue of Microsoft’s Tay chatbot making racist and misogynistic comments.
Third, as shown in Table 1 and Fig. 3, the most frequently mentioned words on online news media and social media were compared. In positive words, “convenient” was the most frequently mentioned followed by “fast” in online news media, whereas on social media “new” was the most frequently mentioned feature followed by “diverse.” The positive words in the two sources were almost similar and most of the extracted words related to functional terms. In negative words, “dry” was the most frequently mentioned word in online news, whereas on social media, it was “worry.” In both sources of online text, the artificial features of chatbot services was in second place by frequency of use. The negative words in online news such as “difficulty” and “rude talk” revealed inconvenience and dissatisfaction with functional aspect of chatbot, while the negative words in social media such as “worry” and “doubt” revealed distrust of the chatbot.
Lastly, as shown in Figs. 4 and 5, each channel showed similar findings compared to Table 1 and Fig. 3. However, there were some cases where the rankings changed in six months. For example, in social media, “new” which was ranked no. 1 in the positive category and “diverse” which was ranked no. 2, came down to 2~3 in half a year. Also, “worry” and “artificial” were reversed. While in online news media, after 1~2 ranking changed in half a year, the ranking in the first and second places remained unchanged after that.
4 Conclusion
These results demonstrate that the number of comments on chatbots has continuously increased over time and revealed a lot of words that were overwhelmingly positive about chatbots. Chatbot services mentioned in online news and social media had the merit of being convenient, fast, smart, useful, and helpful for consumers. Negative words, on the other hand, took up only a small percentage. But the keyword “dry” is what we should pay attention to. Consumers who mentioned chatbot services negatively hope chatbot will be more like humans. In this regard, chatbot services need to be designed to provide not only functional benefits but also emotional and more humanlike values to consumers. Even though chatbot services deliver great values to consumers that are basically based on technological attributes, these functional benefits, on the other hand, may increase consumer resistance to adopt these new technological services. Therefore, service designers should be aware of the fact that consumers’ perceptions on these services are shaped by their cumulative experiences that are affected by not only instrumental aspects but also emotional values.
References
Aditya, D., Alisha, S., Darshana, G., Mrunmayi, D., Prachi, M.: A survey of various chatbot implementation techniques. Int. J. Comput. Eng. Appl. 11, 1–7 (2016)
Hearst, M.: Text data mining: issues, techniques, the relationship to information access. In: Presentation Notes for UW/MS Workshop on Data Mining, pp. 112–117 (1997)
Feldman, R., Dagan, I.: Knowledge discovery in textual databases (KDT). In: Proceedings of the First International Conference on Knowledge Discovery and Data Mining (KDD 1995), Montreal, pp. 112–117 (1995)
Fayyad, U., Shapiro, G., Smyth, P.: From data mining to knowledge discovery: an overview. In: Advances in Knowledge Discovery and Data Mining, pp. 1–36. MIT Press, Cambridge (1996)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2018 Springer International Publishing AG, part of Springer Nature
About this paper
Cite this paper
Jeong, Y., Suk, J., Hong, J., Kim, D., Kim, K.O., Hwang, H. (2018). Text Mining of Online News and Social Data About Chatbot Service. In: Stephanidis, C. (eds) HCI International 2018 – Posters' Extended Abstracts. HCI 2018. Communications in Computer and Information Science, vol 850. Springer, Cham. https://doi.org/10.1007/978-3-319-92270-6_61
Download citation
DOI: https://doi.org/10.1007/978-3-319-92270-6_61
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-92269-0
Online ISBN: 978-3-319-92270-6
eBook Packages: Computer ScienceComputer Science (R0)