NewsSift by Financial Times Talks the Language of Business

Or does it? It's been out for a couple of months, so it's not new news - another expression I always wanted to see in writing. However, it is still interesting for what it could mean to the future of content, now that we have so much of it online - and that of curators as well as creators of that content.

NewsSift is a tool, still in beta, that bases its searches upon relationships and context instead of keywords. For example, if you input the phrase car industry, you will see charts (scroll down on your search) that display sentiment - positive, negative, neutral - article sources - Financial Times shown as a subset of newspapers - and top organizations, places, people and themes. 

The data you get back will allow you to scope what is going on at the moment on the basis of what international news sources are writing about. And yes, blogs are included, although it looks like it's blogs from news organizations so far.

NewsSift has a Twitter stream, and is hiring at the moment. You won't be surprised to see that one of the positions is for content manager. Editors and curators are still critical for the business and for the organization of news in a way that is relevant and meaningful to the person conducting the search. Context, which in this case is how people, places, topics and organizations relate to each other, matters.

Why would the Financial Times go into the search business?

According to CMO of FT Search John Greenleaf, the answer is relatively simple:

"know what you do well; know what you and your brand mean to consumers and advertisers; seek out unoccupied territory in successful growing and profitable markets or segments; and develop a meaningful, sustainable difference and a solid, targeted, efficient business plan to support and grow the effort."

Cross referencing this quote with the job posts, it looks like we're talking about becoming Renaissance marketers -- with experience in content and context, product development and business life cycle, and a solid understanding of the role of digital (media) and analog (social) technology to make it happen. All at the service of meaning - and thus value - for customers.

Instead of looking to be the next Google, FT Search is looking to build on what it already does well and owns, which is creating and archiving content. They're currently in beta, so do take the service for a spin to see what it could do for you. It could help you discover your next market opportunity via the understanding of the semantics of news.

How would you use the service in public relations?

You can research news stories by time frame and sentiment, for example. Or you can look at what is trending up and the type of coverage it has already received. All on the same page. Some ideas for application to attract coverage (you apply these situationally with a foundation in the business):

• go against the grain - use a term or a talk about it in a way that will come up as unique in the sea of conversation around that topic
• go deeper with commentary - ground your story in factual opinion and unleash the power of dialectic
• go wider by defining a category - become the theater where the dialogue is held, create context yourself
• go niche and own one vertical - mine the conversation for the one thing that you can honestly own
• go fast with practical information - if you have expertise that applies to the topic trending up, jump ahead of the conversation
• go for the long tail - if your position is fragmented among many verticals and markets, you might look at which pieces you want to participate in and divvy it up that way.

What else? Would you see an opportunity to have a similar service in other languages, for example? This one seems to search news coverage in English, which is not the language spoken by the largest number of people, although it may be the most influential in business context.

Comments

[Andrew Careaga]

Interesting. I can definitely see the possibilities of this tool for PR purposes, but it looks like the algorithm used to determine positive, negative or neutral coverage could use some tweaking. Just did a keyword search of our institution and the three articles the search tagged as "negative" really weren't. Two were about a researcher's development of a new method to test for cancer, and the third was about an expert's availability to discuss space debris. Perhaps the algorithm screens for "negative" words (such as "cancer") then concludes that the story is negative. If that's the case, then a LOT of positive research stories could end up in the negative category.

[Valeria Maltoni]

That's valuable feedback - I hope FT Search is listening and not just limiting the learning to their own survey on the site. As Google learns daily, algorithms only take you so far. It will be interesting to experience the richness of semantic or people-powered search and filters. Will we have tools that learn? I wrote a couple of posts a while back on Artificial Intelligence agents as Conversation Agents for that reason - imagining what's next in Web technology.

[Josh Chandler]

Wow, suddenly the "power" and "quality" of Google Blog Search, Google News and other "news sources" suddenly become redunant against this news search facility, harbouring every bit of the visual and data based context I am wary of from a "WolframAlpha.com" (on the data side) or "SearchMe.com" (on the visual side) standpoint, but offering unique, and invaluable tools which provides far more superior results then that of any news search facility before it.

I am very excited for the future of Newssift.com, 100% relevant, reliable and verifiable, plus the added addition of the "Sentiment" of the article in the search options, just blows my mind.

[Amy Grabowski]

Many thanks for the Newssift review and feeback.

We are certainly finding that part of our core audience are PR/IR professionals, for the ability to track news stories, mentions of companies or products, and the ability to refine and by sentiment and so we appreciate sharing this POV on Conversation Agent.

Newssift is actively listening to all feedback, from collecting surveys on our site, tracking twitter mentions, reading blog posts and comments, to even closely following the footsteps of how people are using the site. We believe all of this will help us to iterate quickly and often to make Newssift a relevant and useful tool.

In response to how we decipher sentiment, without knowing your query string, I can't specifically speak to your search or your experience. However, if you conduct a keyword search on Newssift for "cancer" and add "positive sentiment" you will see plenty of articles that are indeed positive sentiment.

I also like to show this search as an example of how our sentiment works and results:

Pollution + China + positive sentiment
[http://tinyurl.com/ls3hwn]

In general when you refine by sentiment, you will see some outliers, but it should be generally robust as we are constantly tweaking the annotation to get it as close to perfect as possible.

We are excited to continue building Newssift into a powerful semantic tool to help you SIFT the news and gather insight as part of your daily workflow. And thank you for the feedback and please keep it coming.

[Valeria Maltoni]

@Josh - quality is key. We can now easily do quantity in search, but we are left wanting in many cases on the reliability of the information. This tool is beginning to take things a step further.

@Amy - thank you for jumping in and for helping answer Andrew's question. I like the idea of being able to see outliers and keep refining the search with more definition around terminology. it looks like the tool is based upon a combination of good automated data pull and people analysis.