This is the personal website of Al
yona Medelyan. I am the Chief Research Officer at Pingar
. I lead the development of Natural Language Processing algorithms for tasks such as semantic and faceted search, query analysis, text summarization, keyword extraction, entity and entity relations extraction.
I hold a PhD in Computer Science from the University of Waikato and was supervised by Ian
PhD thesis: Human-competitive automatic topic indexing
I recently gave a keynote at the NZ CSRSC 2013
, a Computer Science PostGraduates conference organized by students for students, which happens every year at a different New Zealand university.
For my talk, I run a survey to find out what happens to Kiwi CS graduates, which resulted in some interesting statistics. My slides and the summary of the results are summarized on my survey page
Last week I spoke at Text Analytics World
in Boston. The slides from my talk are now available on SlideShare
. The talk introduced the area of keyword extraction, how a typical algorithm works and how it can be evaluated meaningfully, and also summarized a use case, where Pingar API
has been installed on an Amazon cloud to extract keyphrases from nearly 2 million publications.
Recently I was interviewed by Sean Golliher at SemanticWeb.com
about Pingar's technology. The interview is one of the first ones in their new innovation spotlight series and focuses on Pingar's keyword and entity extraction algorithms, applying these to unstructured data, including a Twitter-based experiment. I also talk about generating taxonomies on the fly and how Linked Data is used in this research to get great results.
On the 1st of March this year I will be speaking at Strata
, an O'Reilly Media conference with a slogan "Making Data Work". Together with Anna Divoli
, I gave an interview about Big Data and its challenges that is now published on O'Reilly radar
and features on Forbes
. We talk about the practical applications of text analytics and how it can help large organizations that struggle with masses of unstructured data.
For two months this year (September and October) I have relocated to the Silicon Valley, where Pingar
has a new office. During this time I will be attending the SharePoint conference in Anaheim and the HCIR workshop
at Google, Mountain View. Earlier this year I have collaborated with Anna Divoli
from the University of Chicago. Our joint research paper on "Search interface feature evaluation in biosciences
" got accepted for presentation and poster at HCIR. The great news are that Anna will be now joining the Pingar team and relocating to New Zealand.
Staying in California for work is fantastic because it also gave Nathan and me the opportunity to do many fun personal trips: New York, San Francisco, Burning Man, Central California coastal road trip, LA...
I'm stoked that Pingar
is now not only re-branded but also released a set of extremely useful tools out to developers via an API. We provide research-based solutions for query analysis, text summarization, taxonomy mapping, entity and address extraction, text sanitization and profanity checks. The tools are constantly improving and their number is growing. And we even have API for Chinese NLP tools
, which is truly unique.
This month I was lucky to be invited to the Kiwifoo
, a gathering of tech and not so tech people, a wonderful place to get expired and make useful connection. However, what expired me even more, are the discussions in the Kiwifoo forum after the earthquakes in Christchurch. Immediately, participants came up with ideas for organizing accommodation, tech support, hot desk support and social media volunteering, but most importantly they made things happen
! Forum members influenced Trademe to create a support forum
, Google to link a resources page
from their landing page, a Christchurch recovery map
was put together by a team of volunteers. It's great to be a part of such an active community. New Zealand (& overseas) people strive to help so much that donation sites like Red Cross
break from too much traffic (now made working!).
A linguist/philosopher/IT enthusiast from Hungary Zoltan Varju
asked me to answer a few questions for his series of blog posts "The life of a computational linguist". Zoli previously interviewed Oliver Mason
, Jason Adams
and Hugo Liu
. Here are my answers
, which include thoughts on favorite theories & technologies and an honest description of what I do in real life. :-)
Over the last few months I have been porting Pingar's semantic algorithms from English to Chinese language. It was a challenging and interesting experience.
Chinese differs from English significantly, but in terms of extracting query refinement suggestions, extracting keywords and generating summaries, the same general techniques can be applied. A demo of the resulting product is shown in this video
from 1:30. The National Business Review reports about its launch at the Shanghai Expo
I have just updated my publications list
with two new entries. The first one is a demonstration paper which I will present at JCDL in Australia in June. It describes how Maui can be integrated into a Google App web application to support human indexers in their work. The second one is the result of a one year collaboration with Su Nam Kim, Min-Yen Kan & Tim Baldwin. We have been helping Su Nam Kim in setting up an international competition in automatic keyphrase extraction from scientific publications. 19 teams have submitted their results and I look forward reading the descriptions of their algorithms, which will be unvailed at the SemEval workshop in Sweden this July.