With the data ingested, enriched and normalized, IBM® Watson® Discovery Service can begin to help
your users find the needles in their data haystacks. It does this by shifting the paradigm away
from traditional keyword search to a model that more closely aligns with how people think.
This shift means Discovery doesn’t conduct a keyword search, then post-
process the search results to extract the wheat from the chaff. Instead, it automates what has
traditionally been a two to three step process.
So Discovery brings not just straight keyword-based search, but also context
into the results without requiring post processing. As a result, users can quickly see the entities,
the relationships and general, broad categorization like taxonomy.
Because this mirrors how people think, without going through multiple refinement steps, users
can quickly compare the results from different searches on the same topic. For example, a
search for a specific company such as IBM could return the documents that mention
IBM—including the company along with its various products and services.
Built-in connectors allow you to get content from the following systems into your
search index without writing any code:
- The history of IBM
- Key current IBM employees
- Recent patents that IBM has filed
- Companies IBM has acquired
- IBM market valuation over time
- Specific spikes or drops in IBM market valuation
- Events that may have triggered those sharp valuation changes
A query capability built on Watson artificial intelligence, combined with enriched metadata
from structured and unstructured sources and the hierarchical view of that data, make this
possible. And Watson Discovery Service wraps these capabilities up in a query language that
targets the way users actually ask questions to generate deep insights without the noise.
The difficulty is that constructing the plumbing with the API calls and associated coding can be
time consuming and costly. And the learning curve associated with building the code
infrastructure may be daunting. Discovery comes with that plumbing
already built—so these enrichments are ready to use.
Additional enrichment tools
The result—deep insights without requiring users to climb a steep technical learning curve that
unlocks that powerful search. To illustrate how this works, let’s return to our online retailer, Blue
Discovery ingested and enriched the unstructured data from the Blue Snail
Style website along with user reviews and comments. It then combined that with the structured
data from the sales and financial systems in a queryable, hierarchal structure.
Now the marketing department is using the user-centric query language in Discovery
to extract insights that could drive future product decisions, supply chain adjustments
and marketing plans. For example, they discover that:
- Clothing that is not machine washable has the highest profit margins.
- Machine washable products have higher per-product sales volumes than non-machine washable products.
- Clothing from two South American sources has the highest incidence of backorders.
- Women’s products only account for 32 percent of inventory although sales to female
customers make up 52 percent of revenue.
- Customers like coral pink and other lighter shades of red rather than maroon and the darker red tones.
- Customer comments point out Middle Eastern cotton clothing is softer and wears better
than other purchases, which may help explain why it has the lowest rate of returns.
This is just the beginning of what’s possible with the power of cognitive computing. And IBM
Watson Discovery Service makes it easy for you to take advantage of that power without having
to climb a steep learning curve.