NSTM: Real-Time Query-Driven News Overview Composition at Bloomberg

Abstract

Millions of news articles from hundreds of thousands of sources around the globe appear in news aggregators every day. Consuming such a volume of news presents an almost insurmountable challenge. For example, a reader searching on Bloomberg's system for news about the U.K. would find 10,000 articles on a typical day. Apple Inc., the world's most journalistically covered company, garners around 1,800 news articles a day.We realized that a new kind of summarization engine was needed, one that would condense large volumes of news into short, easy to absorb points. The system would filter out noise and duplicates to identify and summarize key news about companies, countries or markets.When given a user query, Bloomberg's solution, Key News Themes (or NSTM), leverages state-of-the-art semantic clustering techniques and novel summarization methods to produce comprehensive, yet concise, digests to dramatically simplify the news consumption process.NSTM is available to hundreds of thousands of readers around the world and serves thousands of requests daily with sub-second latency. At ACL 2020, we will present a demo of NSTM. * Order reflects writing contributions; M.X., I.C.C., and J.B. designed and developed a prototype of the system; All implemented the production system; A.A. managed the project. I.C.C. worked on the project while employed by Bloomberg.

Islander: A Real-Time News Monitoring and Analysis System

“…Watch 'n' Check (Cerone et al, 2020) builds a social media monitoring tool for collaborating with factcheckers to detect misinformation online by assisting them in the identification and targeting of misleading claims. NSTM (Bambrick et al, 2020) provides a summarization engine that condenses large volumes of news into short, easy-to-absorb points. The engine helps users digesting the key news about companies or countries.…”

ClioQuery: Interactive Query-Oriented Text Analytics for Comprehensive Investigation of Historical News Archives

“…(This is sometimes described as extractive summarization [33] because the output text is extracted from input text.) User-facing systems such as Newsblaster [89] and NSTM [3] apply this research by showing such textual summaries in a graphical interface. We say that such tools implement the textual summary design pattern (Figure 3b).…”

“…Showing a single document in a single window or tab is a common interface pattern. It is employed, for instance, in the Overview[13] and Jigsaw[123] document viewers, and in traditional search user interfaces, which often link to individual documents from a main search engine results page[29, Section 6.3] 3. Clio is a common prefix (e.g., ClioVis[15]), implying a connection with history.…”

News Headline Grouping as a Challenging NLU Task

“…Several of the most-used news aggregators, such as Yahoo News 2 , Google News 3 , and Bloomberg's NSTM (Bambrick et al, 2020) present headlines in groups. As these systems do not have published algorithms, we cannot comment on their methods; nonetheless we hope that the release of the HLG dataset offers a common evaluation test-bed to benchmark systems.…”