Good debt or bad debt: Detecting semantic orientations in economic texts

Abstract

The use of robo-readers to analyze news texts is an emerging technology trend in computational finance. Recent research has developed sophisticated financial polarity lexicons for investigating how financial sentiments relate to future company performance. However, based on experience from fields that commonly analyze sentiment, it is well known that the overall semantic orientation of a sentence may differ from that of individual words. This article investigates how semantic orientations can be better detected in financial and economic news by accommodating the overall phrasestructure information and domain-specific use of language. Our three main contributions are the following: (a) a human-annotated finance phrase bank that can be used for training and evaluating alternative models; (b) a technique to enhance financial lexicons with attributes that help to identify expected direction of events that affect sentiment; and (c) a linearized phrase-structure model for detecting contextual semantic orientations in economic texts. The relevance of the newly added lexicon features and the benefit of using the proposed learning algorithm are demonstrated in a comparative study against general sentiment models as well as the popular word frequency models used in recent financial studies. The proposed framework is parsimonious and avoids the explosion in feature space caused by the use of conventional n-gram features.

Sentiment-Driven Stochastic Volatility Model: A High-Frequency Textual Tool for Economists

“…The human-annotated financial phrase bank, constructed by Malo et al (2014), can be used for training and evaluating alternative models for financial and economic news texts. Having this training data, we are able to calibrate the scoring function s(X) in (1).…”

SemEval-2017 Task 5: Fine-Grained Sentiment Analysis on Financial Microblogs and News

“…• Yelp Restaurant Sentiment Lexicon -adopted by Chen et al (2017) • Amazon Laptop Sentiment Lexicon -adopted by Chen et al (2017) • Macquarie Semantic Orientation Lexiconadopted by Chen et al (2017) • Harvard's General Inquirer Lexicon -adopted by Nasim (2017) • IMDB -adopted by Jiang et al (2017) • AFINN -adopted by Jiang et al (2017) • DepecheMood Affective Lexicon (Staiano and Guerini, 2014) -adopted by Mansar et al (2017) • Amazon Product Reviews 16 -adopted by John and Vechtomova (2017) • Financial Phrasebank (Malo et al, 2014a) -adopted by John and Vechtomova (2017) • Corpus of Business News -adopted by Pivovarova et al (2017) In total, four lexica listed above are the ones mostly used (all by 4 participants each): (i) the Loughran and McDonald Sentiment Word, (ii) SentiWordNet, (iii) Opinion Lexicon and (iv) Harvard's General Inquirer Lexicon. Unlike the case in track 1, none of the participants ranked first till third used one of these four lexica.…”

“…Furthermore, sentiment scoring will be more fine-grained as it will consist of floating-point numbers in the range of -1 (very negative/bearish) and 1 (very positive/bullish), with 0 representing neutral sentiment. (Malo et al, 2014b) present the Financial Phrase Bank, a resource containing around 5000 sentences from English-language news about companies listed on the Helsinki stock exchange. Annotations at the level of syntactic phrases assigned one of three sentiment classes (positive, negative, neutral), based on the expected influence on the stock price.…”

Assessing News Contagion in Finance

“…Despite the good results, there are applications where it could be preferable to avoid dictionaries in favour of more data driven methods, which have the advantage of higher data coverage and capability of going beyond single word sentiment expression. Malo et al (2014) provide an example of a more sophisticated supervised corpus-based approach, in which they apply a framework modeling financial sentiment expressions by a custom dataset of annotated phrases. In the last years, different papers, embracing the data driven approach, have used the deep learning models to analyze textual data and have shown good results in predicting distress events of financial institutions (Rönnqvist and Sarlin 2017;Cerchiello et al 2017a) and in predicting S&P500 stocks (Ding et al 2015).…”