How We Do It
Three things lie at the heart of Topsy’s platform – a realtime counting and metrics machine; a dynamically-sorted index of the social web; and a data enrichment system. Together, these technologies form one of the largest “big data” projects; we’ve indexed over 100 billion items.
By counting, indexing and enriching Twitter and social web data in realtime at unparalleled scale, Topsy is able to delivering unique technological innovations that power the world’s most powerful social analytics.
-
Topsy Influence
Topsy Influence measures the likelihood that others will pay attention when you say something. For each author, we analyze all posts and attention they receive from other people: attention from people with influence matters more. Topsy Influence powers many of our other capabilities, such as relevance ranking, experts and trending topics.
For more information about Topsy Influence, read our whitepaper here.
-
Comprehensive & Exact Counts
Topsy gives you the unique ability to instantly get the number of mentions (by minute, hour or day) for any term, phrase, username, link or hashtag – up to the current minute. No estimates, no sub-samples. Comprehensive.
-
Relevance Ranking
Topsy’s Relevance Ranking surfaces the most important content on the social web, for any term, for any time period – the past few minutes, or a couple of years ago. It uses Topsy’s realtime index, URL resolution, baseline metrics and influence weighting, among other factors, to ensure that valuable signal is extracted and spam is removed.
-
Geo-Inference
Since only 1% of tweets are explicitly geotagged, Topsy developed machine learning methods to infer an author’s location, using features such as regular references to landmarks or events. By doing so, we have high-confidence geographic information for more than 90% of tweets.
-
Sentiment Analysis
Topsy Social Sentiment is computed for millions of terms, based on hundreds of millions of tweets every day. It is tailored to the informal and abbreviated language that is often found in tweets and social media in general, and normalized for each term based on scores for all other terms used in Twitter.
For more information about Topsy Social Sentiment, read our whitepaper here.
-
Related Terms Discovery
You don’t know what you don’t know: but given a set of terms you are already tracking, Topsy algorithmically helps you discover new terms, phrases, accounts and hashtags that are related to and trending along with your topic.
-
URL Resolution
Topsy expands every link posted on Twitter to its full, final form, following through all URL shorteners and website specific URL redirects. This means that Topsy always has canonical counts of references to a particular piece of content shared, regardless of the type of content, such as tweets, web pages, photos, and videos.
-
Cumulative Exposure
Topsy measures the total number of times a term or hashtag has appeared in all Twitter timelines, and tracks this minute by minute. Together with Topsy’s precedence and amplifier analysis, this allows you to identify how terms, hashtags, or other content is resonating with users, and track the effectiveness and reach of a campaign.
For a case study using cumulative exposure, read our whitepaper here.