Harnessing the “wisdom of employees” from online reviews

Assessment of Luxury Trains in India: A Case Study of Maharajas’ Express

“…Use of probabilistic topic modelling method was ideal given the nature of this study (Stamolampros et al, 2019a;Stamolampros et al, 2019b). "In principle, topic modelling is a set of unsupervised machine learning techniques which self-organize textual corpora in groups of topics evaluating how specific groups of words appear together using both volume and context as inputs" (Stamolampros et al, 2019b, p. 18).…”

“…"In principle, topic modelling is a set of unsupervised machine learning techniques which self-organize textual corpora in groups of topics evaluating how specific groups of words appear together using both volume and context as inputs" (Stamolampros et al, 2019b, p. 18). Similar to studies of comparable nature (Roberts et al, 2014a;Stamolampros et al, 2019a), this study made use of the Structural Topic Model (STM), a generative model of word counts (Roberts et al, 2014b) that allows the inclusion of document metadata. To estimate the data using Structural Topic Model (STM), the following preprocessing steps were undertaken (a) word text tokenization (splitting of text into a list of tokens) (b) standardization (conversion of characters into lower case) (c) removal of numbers and punctuation marks (d) removal of stop words and contextspecific words and (e) stemming (reducing inflected words to their root forms).…”

“…Infrequent terms that appeared less than 3 times in the corpus were removed as part of the preparation process (Roberts et al, 2014b). The number of topics (K = 3) were determined based on (a) held-out likelihood (b) exclusivity of topic words to the topic and (c) semantic coherence of the topic structure (Stamolampros et al, 2019a;Stamolampros et al, 2019b). The topics were then labelled with the help of a luxury train supervisor, who accompanied the authors throughout the interviews.…”

Online Employer Reviews as a Data Source: A Systematic Literature Review

“…So far, the emergence of product and service reviews has produced an extensive body of research [see, e.g., 21]. Since employer reviews likely provide rich information beyond that obtained from product and service reviews [see 41], employer reviews open up a rich and novel field of research.…”

“…For instance, the employer review website Glassdoor reported a database of 55 million employer reviews, CEO approval ratings, salary reports, and other job insights in May 2020 [15]. The emergence of online employer reviews, as a unique type of user-generated content, is likely to provide new research opportunities beyond the work undertaken so far concerning online reviews of customers on products or services [41]. Harnessing these opportunities requires that researchers can identify the research topics addressable with employer review data and the information that can be extracted from these reviews' content.…”