PT - JOURNAL ARTICLE AU - Julia Klevak AU - Joshua Livnat AU - Kate Suslava TI - A Practical Approach to Advanced Text Mining in Finance AID - 10.3905/jfds.2019.1.1.122 DP - 2019 Jan 31 TA - The Journal of Financial Data Science PG - 122--129 VI - 1 IP - 1 4099 - https://pm-research.com/content/1/1/122.short 4100 - https://pm-research.com/content/1/1/122.full AB - The purpose of the study is to illustrate one application of unstructured data analysis in finance: the scoring of a text document based on its tone (sentiment) and specific events that are important for the end user. The methodology begins with the well-known practice of counting positive and negative words and progresses to illustrate the construction of relevant events. The authors show how the application of this methodology to the analysis of earnings conference call transcripts produces a signal that is incrementally additive to earnings surprises and the short-term returns around the earnings announcement. An interesting feature of the tone change extracted from the conference calls is that it has a relatively low correlation with both earnings surprises and the short-term return around the earnings announcement. This indicates how use of text mining and scoring of unstructured data can add information to investors beyond structured data.TOPICS: Big data/machine learning, fundamental equity analysis