Housing Media Analysis, Using IBM Watson’s Natural Language Processing

Housing Media Analysis Sentiment ScoreHousing Tides Media Analysis Sentiment Score Article Chart

IBM Watson natural language processing technology allows Housing Tides to correctly understand and synthesize large volumes of housing media, including translating language into “sentiment” or the author’s explicit meaning, and aggregate that information in a consolidated form that users can easily access and understand. Watson AlchemyLanguage Relation Extraction enables quick identification of forecasts offered by industry experts in housing news, which we assess to glean insights about whose forecasts are most accurate. Watson Natural Language Classifiers provide a reliable filter so that the Housing Tides team can focus our analyses on news relevant to the housing domain.

The result is a gauge of the overall media sentiment surrounding the housing market and home construction industry. Sentiment scores are determined for each news article or blog via the IBM Watson Alchemy Sentiment Analysis API, using natural language processing to assign a sentiment value to each piece of text. Sentiment scores range from -1 to +1; articles expressing the most negative sentiment earn a -1 and the articles expressing the most positive sentiment earn a +1. Most articles fall within a narrow band as the majority of news expresses mixed positive and negative sentiment.

With Watson Logo

Share this Post