Whether you are looking for people to fill a vacancy, or for vacancies that you might be able to fill with candidates from your own resources, the likelihood is that there'll be a lot of time-consuming and inefficient donkey-work before you reach a successful outcome. The main problem is that Internet search sites - while they've come on a long way - are still not very clever.
Nine out of ten Internet sessions begin with a search. But a lot of those searches end in frustration. Let's say you are looking for a hotel holiday in the South of France. You aren't really too bothered about the exact location, but want a pool and childcare facilities. So you put all that into a search engine:
"hotel South France pool childcare".
What comes back is a mess. More than 34,000 results - some of them from individual hotels, most from major holiday chains. You can't tell at all which are the most popular or reasonably priced. You start digging through the top results. A week later, you still don't know whether your shortlist is missing some of the best available options.
The trouble is that the search engines are looking for the words "hotel South France pool childcare" and every single holiday provider in the world with have pages containing those words. The ones who come top are the ones from the best-known companies, or the ones that have done the best job of optimising their websites to suit the way search engines work. The list might include hotels which "do not have childcare" - but since they've used the word 'childcare' on their page, they get added to the results anyway.
What many people have begun to think - including the inventor of the Web, Sir Tim Berners-Lee - is that search needs to be rethought from the ground-up.
In 2001, Berners-Lee wrote about the development of a new way of finding information on the Internet:
"The Semantic Web will bring structure to the meaningful content of Web pages, creating an environment where software agents roaming from page to page can readily carry out sophisticated tasks for users."
'Semantic' means looking at the meaning of the words on the Web pages. A web search based on this technology would understand that 'no childcare' shouldn't be included in the results for a search of hotels with childcare. In a recruitment scenario, it would know that someone who wrote in their CV that they 'reported to the senior project manager' is not a senior project manager themselves. More sophisticated applications would be able to match every element of a person specification and eliminate candidates who did not meet all of the essential criteria, ranking the remainder by how many of the desired criteria they possessed.
Daxtra tries to make its products work in the same way recruitment companies do. Its core offering, Daxtra Capture, takes CVs, works out what they contain and puts those contents into your database. This means that employees aren't wasting time retyping information, but rather can attend to tasks that computers can't yet do.
This is a little bit more complicated than it sounds. For example, if a CV has the words 'Paris Hilton' in it somewhere, how does the computer know whether that's a hotel or a celebrity?
Candidate Capture employs a Natural Language Processing technology as well as a suite of rules and statistical inferences to make a decision. In this case, it might look to see whether the words appear in a 'Previous Experience' section or 'Hobbies and Interests'. It would look for clues like whether 'France' is mentioned nearby. It would know that it's highly unlikely, though not impossible, that the candidate's name is Paris Hilton and look for the contextual and semantic evidence to decide exactly how to categorise this piece of information.
Another example would be the word 'Director'. The CV might be from someone who had been a director at a company, someone who's directed a film or a play, or perhaps might be indicating proficiency with Adobe's Director software package. A system that relied on looking for keywords would never know which was intended. Only by looking at the word in context and by understanding the structure of the whole document can the word be understood correctly.
Andrei Mikheev, founder and CEO of DaXtra, gives the example of referees listed on a CV. " If the program simply looked for anything that looked like a name and address, then how would it know that the referee's details aren't the candidates?"
Daxtra's latest product, Daxtra Search was developed to help recruiters solve some of the headaches of Internet search.
The landscape has changed and the ideal candidates may not be on the recruiter's database, but rather be using online job sites like Monster, or social networking sites like LinkedIn and Facebook to indicate their availability. Of course, each of those job boards and networking sites lays out the information in a different way, may or may not allow Google and other search engines to index its content and may require logging into the site before any searching can be done. The result of this proliferation of recruitment vehicles could be both time-consuming and costly.
Daxtra Search works a specialised, semantic search engine tailored to matching candidates with vacancies. It automatically logs into all the job boards and networking sites you are interested in and conducts simultaneous searches on all of them. Again, it employs all of the language processing technology to obtain the best fits - someone who listed themselves as a 'programmer', for example, would still be included if your search was for a 'developer', but excluded again if the evidence suggests that they developed film or a series of events.
Mikheev believes that Daxtra's approach to search could be expanded upon to help in other professional domains and in mainstream web search. Today, for example, if you search for a cinema in Reading, you do get film results, but you also get lists of suggested 'reading' matter about cinemas.
DaXtra's technology could already work to cluster those results so that it's quicker to get to the results for the meaning that you wanted. If you are searching for a certain hotel in Paris, once more, then being able to quickly remove all the results about that celebrity would make life a lot easier.
There are also moves to better integrate search. The divide between offline information stored in your database and online information stored on social networking sites is also being dissolved. Information you store offline will be increasingly linked and spidered without you needing to make any effort whatsoever. Let's say you are looking at information about someone who used to work for IBM or who has a particular qualification, and you wonder who else you know with those characteristics. You could launch a database query, of course - but that will only find the people in your database. Integrated search would not only find people you've already registered, but also people you or your colleagues have connected with on Linked-In, Facebook and other Internet sites. Suddenly, half-a-dozen or more search queries and their results become one.
As the amount of information available online continues to increase at an exponential rate, we all need better ways of getting to what it is we need.