Explaining semantic search (in short and not-so-short)
This is a question I get quite often: what is semantic search? I’ll give you the short and not-so-short answer. As well as some possible applications which we’ve included our own semantic search platform.
Semantic search in short
In short: semantic search extends the current search possibilities by looking at 1) the intent of the searcher as well as 2) the context within which the search takes place. When properly combined, you get more refined and better attuned search results.
OK, that makes sense. But when you do this on the general web, the task of uncovering intent and context becomes quite challenging. Imagine you’re planning a business trip to Paris. You go to Google and type in the keywords ‘Paris’ + ‘Hilton’, only to get dozens of celebrity pages. You’ll have to refine your keyword string (no pun intended!) to get to the actual Paris branch of the hotel chain.
And yes, Google is already making some assumptions and offers you some additional keywords to try and understand your intention better, but much of that is still after the fact.
Semantic search, not-so-short
Semantic search becomes a lot more actionable when you place it in a pre-defined context, such as an industry or a company. Instead of offering ‘best guesses’, the search technology can more quickly make assumptions about intent and context and fine tune the results accordingly.
Consider an R&D-driven industry like life sciences. What we do there, is connect different data sources (open and closed public data, internal and third party data) and tie it all together. We create an ontology, by which we mean that different descriptions for the same thing are given a unique identifier. So the system knows that ‘male’ is the same as ‘m’, is the same as ‘masculine’, and so on. That way, data from an unlimited number of different data sources can be compared and meaningful results can be generated. The strength of semantic search is that it can do this for an unlimited number of categories.
Semantic search and Big Data
In this way, semantic search overcomes the 4 Vs of Big Data: volume, velocity, variety and veracity. Making connections between ever-growing mountains of data and widely differing data sources is fueling the need for semantic search.
With semantic search, it doesn’t matter how many different databases you search in. The system will make the corresponding links between the bits of information and present it in a meaningful manner to you.
Semantic search reveals unexpected connections between your search query and different data sources. Consequently, it retrieves data you didn’t know you were looking for. Unlike regular search, which is mostly descriptive, semantic search can be both predictive and prescriptive. Semantic search makes you a whole lot smarter.
Semantic search also uncovers interconnections and relationships between data. Data is combined, compiled and represented in a meaningful way. Your data is enriched, making it more valuable.
Semantic search applications
Here are some cool items we’ve added to our own DISQOVER semantic search platform:
- Saved search queries and collaboration: name your search, save it and share it with colleagues. Others can rerun your search, extend it or choose a different path, somewhere along the line, in order to uncover alternative routes for discovery. Collaborating with others on search patterns becomes a lot easier.
- Rerun your saved searches. Because data changes all of the time, your research findings will change too. With saved searches, any new data since your last search, will automatically be included in your rerun. That way, you no longer miss out on relevant updates. The saved search path lets you revisit your search on a later date and rerun, resume or change it as you see fit.
- Visualize data. The big bang of big data makes deriving insights hard. Visualization helps to better understand search results and facilitates telling a better data story. As your search progresses, our DISQOVER platform automatically updates the search result visualizations while new search criteria are offered for additional refinement. By simply adding or removing search filters, you can zoom in or out on the findings.
- Depict the source. Instead of just giving you a search result list, we also create compiled content pages with data from different sources. But because it may be worthwhile to know which data comes from which source, we color code each part. That way, as a (re)searcher, you can define which information is more relevant or reliable, or which sources you would like to research further.
“The Semantic Web is not a separate Web but an extension of the current one, in which information is given well-defined meaning, better enabling computers and people to work in cooperation.” Tim Berners Lee
Everybody becomes a data scientist
Semantic search will become a lot more smarter soon. With ever more data becoming available—including also sensory data of where you are and what you’re state of mind is—generating useful search results will continually become more refined. Until then, companies can already create the right context and generate big wins with semantic search. Semantic search makes complex searches available to all, turning everyone into a data scientist.