Case Study – The Weather Channel®
Only at The Weather Channel – 15 Words for Rain
The people at The Weather Channel® (TWC) have a word for rain – in fact they have more than a dozen words for rain – drizzle, mist, sprinkle, etc. They’re in the weather business so that’s a good thing.
It’s not a good thing to have a lot of different terms for the same thing when you’re trying to find everything having to do with a weather situation, like storm preparation or cleanup. And it’s especially not a good thing when your job consists of logging, storing, and retrieving video and still photos about weather.
With a Library that includes nearly 18,000 media assets totaling hundreds of hours of weather-related video, a consistent, dependable way to log and describe those assets is vital. Particularly when there are more than a dozen terms used to describe just one type of precipitation – rain!
In the past, a video producer putting together a video about storm damage could ask for a clip of an 18-wheeler overturned by a tornado. To be thorough, because the clips of his “18-wheeler” had been logged in as a “semi” or “tractor trailer,” the TWC staff may have had to perform multiple searches to find exactly the one the producer requested. Since taking a long time to find the requested videos or hiring more staff simply weren’t options, TWC staff decided to purchase Data Harmony’s Machine Aided Indexer (M.A.I.) to solve their logging/finding of video problem – the hardest part of their jobs.
M.A.I.™ Changes Search to Found at The Weather Channel
Now, using M.A.I., a TWC employee logs in a clip with an overall title like “Greensburg Kansas Tornado” and a subtitle of “cleanup” or “aftermath.”But it’s the synopsis field that really separates the clip from all the other tornado cleanup clips. A description of what the TWC staffer sees on the video is entered in natural language – “large pile of debris,” “closeup of chainsaw,” “cutting fallen branches,” etc.
Then the M.A.I. Lookup button is clicked bringing up a list of suggested keywords that were selected using rules like “near” or “with,” as in, if “chainsaw” appears near “fallen branches,” M.A.I. will return “storm cleanup” in the list of keywords. As for the 18-wheeler problem, now when that term is entered at TWC, the 18-wheeler is either accepted or the user is redirected to use “semi” or “tractor trailer” or “big truck,” whatever has been agreed upon as the preferred usage.
“We used to have to enter data into the keywords field manually, after writing the synopsis. Now, we simply click the ‘M.A.I. Lookup’ button, and the keywords field instantly auto-populates with the keywords generated by M.A.I.,” explained Jay Tellock, Librarian, The Weather Channel.
The keywords come from a list of preferred synonyms to point to with all the aliases and variations combined into a single rule. All the variations on the term are captured, making it easy for input to the system as well as retrieval by the searcher. No matter what term the requester uses for their question the system will return all the relevant results.
Jay added, “Saving time has clearly been the biggest benefit. Searching for ‘storm preps’ instead of ‘boarding up,’ ‘sandbagging,’ ‘hurricane shutters,’ ‘evacuation kits’—the list goes on, probably cuts search time by as much as 50%, depending upon other unique search criteria for a specific request.”
Since implementing M.A.I. the search takes a few minutes and the results are exactly what was asked for. Findability is increased, saving staff time and minimizing frustration.
Joyce Jefferson, Manager Library Services, The Weather Channel, is very pleased with the performance of DataHarmony’s Machine Aided Indexer:“Prior to M.A.I. we did not have a well-controlled vocabulary; the indexing was loose and unreliable. In the absence of an automated index of terms, each of our four Librarians would use his/her own selected terms to describe scenes. Because of the existence of multiple terms describing similar scenes, our old search process required several steps.
“In order to identify all possible information that would satisfy a request, an average of seven searches was conducted for each search request. M.A.I. has improved the accuracy of information retrieval and eliminated duplicated efforts, thus improving productivity by processing more material in the same amount of time,” she added.