How Our AI Extracts the Most Impactful Commentary from Transcripts

Prashant Budania


March 26, 2021

A critical aspect of researching a company is dissecting and understanding their earnings call and identifying the essential and relevant commentary. Analyzing inflection points in the sentiment of earnings calls has long been regarded as a proxy signal for stock movement and shifts in corporate strategy. Unfortunately, while humans are incredibly good at distinguishing between positive and negative sentiment, it takes a long time to undertake this analysis over several documents. Moreover, as more and more documents are published, the burden to analyze the sentiment of documents only continues to increase.

The AlphaSense platform looks to solve this problem by making it easy to discover the most critical viewpoints and statements, removing the burden of sentiment analysis from the user. In this way, the platform can analyze sentiment across many transcripts and show users the sentiment trends and the positive or negative statements within the transcripts themselves. In other words, it helps users quickly understand the critical commentary (positive or negative) being expressed.

An AI model that understands natural language is needed to accurately and quickly perform this mass sentiment analysis. Our AI allows AlphaSense to understand what someone means when they use a word or phrase (i.e., the sentiment behind the language) within the context of the more prominent statement.

AlphaSense’s sentiment analysis for transcripts uses natural language processing (NLP) and machine learning to parse the language in those documents, revealing the essential snippets and significant changes since previous quarters, down to the sentence level. 

Our sentiment model is built on the bleeding edge of advancements in Deep Learning and NLP. What makes the AlphaSense model so advanced is our use of Transformer-based language modeling architecture to produce state-of-the-art representations of language and context. 

Machine learning models, powerful as they are, require a lot of labeled data to do well on specific tasks. And frequently, labeled data is scarce or too resource-intensive to produce, while raw and unlabeled data are abundant. Language Modeling enables us to leverage the raw data, without needing human-labeled data, to learn powerful representations of the language. 

AlphaSense Language Model

As part of our research work, we have developed a proprietary AlphaSense Language Model (ASLM) trained on over 8 million financial and business documents, giving it a cohesive understanding of the language used and its underlying concepts. Since AlphaSense’s Language Model is trained on large amounts of financial and business data, it can be fine-tuned on different NLP tasks and data sources, meaning it requires less human-labeled data for each new content source or industry vertical while taking advantage of the massive amounts of labeled data we’ve already produced. We have already started to leverage our AlphaSense Language Model for entity tagging, themes, and KPI extraction and have seen encouraging results on our other research projects.

For analyzing sentiment, the AlphaSense Language Model leverages our proprietary human-labeled training dataset, meticulously built over the last ten years. As a result, the Language Model allows us to capture even richer representations of the language and improves the performance of our sentiment model on sentences involving complex language and mixed sentiments. 

Q3 2020 Earnings Call Transcript for Cleaveland-Cliffs Inc.

For example: in the image above, our sentiment model can understand that the statement is a positive signal for the company as they can benefit from it, not a negative one because of the words used (“the pandemic” and “most challenging”).

AlphaSense Language model breakdown

Besides high accuracy, another critical requirement for our sentiment model is interpretability, i.e., allowing users to quickly identify important phrases and sentences driving the model’s predictions, so users build trust in its predictions. Therefore, we developed a novel method to add the interpretability component to our underlying Language Model, highlighting the crucial phrases without adding any overhead to the model’s processing speed. 

 Teledyne Technologies Inc Q4 2020 Earnings Call

The AlphaSense platform offers users the industry’s most accurate sentiment-driven analysis via our proprietary Language Model. For users, identifying and scoring sentiment is essential for spotting critical insights, ultimately generating ideas, and understanding markets and companies.

If you’re an AlphaSense client or trial user, you can get started with sentiment analysis on the platform. If you’re not yet a client, you can request a free trial.

Share this article

Prashant Budania

Prashant Budania is a Senior AI Research Engineer at AlphaSense where he works to solve interpretability, language modeling, and sequence classification problems. He is a Carnegie Mellon and Indian Institute of Technology Delhi graduate with a degree in Electrical and Computer Engineering.

More like this