That also illustrates that in order to do data science effectively you require some domain knowledge to quantify the value of what you have found, he says. “Using the scientific method, you need to know ahead of time what is the likely value if your hypothesis is proven true and then how such insight will influence decision making. The insights have to be actionable to realise the value. That all comes from domain knowledge.”
Industry led
That is also the thinking that underpins the objectives and work of CeADAR, the UCD-based Centre for Applied Data Analytics Research established in late 2012. “The key point about our work is that it is industry-led,” says Centre director Dr Edward McDonnell. “The research proposals come to us from our industry consortium partners and the choices of which to pursue are made by a joint committee according to our three research themes — Intelligent Analytic Interfaces, Data Management for Analytics and Advanced Analytics. The most important metric for us is the transfer of our research outputs back to industry.”
The research proposals come to us from our industry consortium partners and the choices of which to pursue are made by a joint committee according to our research themes. The most important metric for us is the transfer of our research outputs back to industry, Edward McDonnell, CeADAR
CeADAR’s academic partners are DIT and UCC as well as UCD and it is supported by Enterprise Ireland and IDA Ireland. Perhaps surprisingly, its industry partners are mostly Irish enterprises, predominantly SMEs, and just 20% multinationals with operations here. It is also unique for such an institution in its quickfire business approach. Projects last for just six months with tightly defined objectives and aim to produce practical, usable results, Dr McDonnell says. There can be up to 20 projects on the go simultaneously, so that even in its short existence CeADAR has built up an impressive back catalogue.
The range of smart technologies involved ranges from artificial intelligence and machine learning through text-so-speech/speech-to-text and machine translation to data visualisation. For the most part they are employed to deal with large volumes of unstructured data. The results of the CeADAR projects are usable software tools, usually in specialist niches, that are on a par with or indeed indistinguishable from commercial products. TopicListener, for example, uses automatic speech recognition to enable tracking of topics across contact centre call traffic and online video transcript data. Developed for contact centres, it can be extended across other large volume speech audio domains. Another example is IdentityMatch, which identifies persons of interest in social networks. That could be key influencers but is also applicable to searches for people with specific expertise or experience, in certain locations or business sectors.
Another called NudgeAlong is right in there where marketing and analytics are coming together. It extends form customer monitoring through personalised purchase incentives and their design to a final stage of personalised communications. NudgeALong is a recommendation engine that uses all of the information built up about the specific customer, from purchase history, outcomes of previous communications and preferred channels all the way to generating actual message content.
Subscribers 0
Fans 0
Followers 0
Followers