Time Explorer draws its information from a collection of 1.8 million New York Times (NYT) articles dating from 1987 to 2007. With its ability to recognize references to specific times in the future – “June 2012” for example – or use the article’s publication date to determine relative timings – “merger next year” – the search engine is able to show the coverage relevant to a search term over time. This is similar to Recorded Future’s “momentum,” indicating when the search term has attracted the most heat in terms of articles.
Brief summaries of the articles themselves are listed below the timeline, while the titles of articles falling within a particular year – indicated by a scrollable blue highlight – appear above the timeline. Meanwhile, a box to the right displays the names of people, organizations and places that feature prominently in the articles. These can be added to the timeline with a click to compare the coverage of both topics.
Time Explorer was among a number of ideas for using the same NYT dataset presented at a session of the Human Computer Interaction and Information Retrieval (HCIR) workshop. It won the most votes from attendees for use of the NYT articles and attracted praise from Daniel Tunkelang, a tech lead at Google’s New York office who chaired the session.
Although Time Explorer is only a prototype and currently only works with old news, Michael Matthews, a member of the Yahoo research team, says the service would be best used to give context to a breaking story. Users would be able to see the history of the topic and past predictions made about it.
Matthews and his colleagues at Yahoo’s Barcelona research lab are now in the process of adding more sources, including more up-to-date news sources, blogs and other sites, to the service.