Print
Email
Print Edition Stories
Stuart Garfield
In their own words: David Chang, co-founder/VP of marketing, Snapmylife.com
MHT's Jim Connolly: Social network sites focus on niches, sports
MHT Editor Doug Banks on NECN: How companies are using social media
Some day, your computer will know what you want better than you do. Semantic search, the promise behind the buzzword Web 3.0, will find what you’re really looking for, even if you pick the wrong keyword.
Startups have launched to handle this problem specifically, but some of Boston’s other software companies — social web, market research and business intelligence firms, for example — have quietly become centers of innovation behind the semantics scene as engineers work on analytics dashboards that can understand opinion and sentiments.
Communispace Corp. in Watertown, Invoke Solutions Inc. in Waltham, Crimson Hexagon Inc. in Cambridge, Mzinga Inc. in Burlington and Metatomix Inc. in Dedham all see semantic search as critical to their offerings. Yet few of these companies can agree on how a computer program should distill data out of unstructured content.
Communispace calls its analytics suite “premium text analysis.” The company relies on teams of editors to monitor the communities it hosts for its large corporate clients, but they need the software’s assistance to comb through data.
The application goes beyond word count, said Julie Wittes Schlack, senior VP of innovation and design. “Word count is one-dimensional, and you need a big data set,” she said. Their tech is designed to diagram sentences.
Using a library of taxonomies and classifications, the application identifies word combinations that express emotions, including optimism, skepticism, joy and insight, Wittes Schlack said.
Skepticism could describe Crimson Hexagon engineers’ emotions toward sentence diagramming. “Things like (grammar) tend not to matter as much as you might think,” said Chris Bingham, vice president of engineering at the Cambridge company, which monitors online public opinion for clients. Instead, the application compares online content with sample posts, looking for patterns of text, not specific words.
“When people talk, they kind of say the same thing over and over again,” said Candace Fleming, company co-founder and CEO. “Or they back up their point with other information which can be a strong indicator in and of itself of the main point they are talking about.”
Social media marketing company Mzinga thinks the most important data are the connections and activity among users. Mzinga is developing an analytics capability, for release later this month, designed to identify the most important blog posts, comments and threads for companies to watch.
“If you wrote a blog post, and lots of people commented or friended you as a result, that’s useful for the person running that community to know,” said Chip Matthes, senior VP of advanced technologies at Mzinga. The application under development is more in-depth than blog-ranking sites like Technorati, because it considers 70 metrics, not just links and page views, he said.
For now, the tool will be applied to clients’ hosted communities — but mZinga’s search tool could have wider applications, he said. “We built this with the assumption that the initial data we come up with will come out of our communities but ultimately things will span out across the Internet.”
RSS Feeds
Comments
Please Login/Register to post comments.
No comments have been added or approved.