As every year, this past November saw the election of new members to the Speech and Language Technical Committee to take the places of those whose 3-year terms are coming to an end. SLTC membership is grouped by subject area, and in this year’s election (for the 2017-2019 term) we had 17 openings in 8 areas. Current members are eligible to run for a consecutive term once. Before giving the results, let me briefly describe the process of the election itself, since this is one of the essential functions of the committee, and it is important that the process be transparent and well-understood.
The election begins with a call for nominations, published in the late summer or early fall via the SLTC newsletter, as well as other speech and language related forums. Nominations consist of short bios of the candidates and can be submitted by the candidates themselves or a referring individual (as long as the candidate has indicated willingness to serve). A candidate must indicate which technical area he or she is running for. The nomination process closed on November 15, and the full Committee then had until November 30 to cast their votes, via a Surveymonkey page. Votes consist of a ranking of all the candidates in each technical area. Members who are running for re-election cannot vote in the area they are running in. The votes (rankings) are then aggregated as in a round-robin tournament: for each pair of candidates (A,B) we count a “win” for A if A got ranked higher than B more often than not (it counts as a “tie” if A was ranked higher an equal number of times as B). The candidates (again, separately in each area) are then ranked overall by their total wins (where ties count half of a win). The top N candidates are chosen for the available N openings in an area.
The election process is cumbersome because it is stitched together out of home-grown solutions with human intervention required at the seams. Until recently, the nominations had to be manually reformatted into a structured document for distribution to the voters; this is now replaced by entering nominations on a web form that generates a structured database of the candidates’ information. Setting up the vote in Surveymonkey is still a manual process. The vote outcomes are then downloaded, and reformatted and tallied via perl scripts. Clearly it would be preferable if a more standardized, integrated solution for the entire work-flow were available.
This year 49 nominations were received for the 17 openings. The elected candidates with their respective technical areas were as follows, with newly elected ones highlighted:
Milica Gasic (Dialogue Systems); Eric Fosler-Lussier (Natural Language Processing); Fred Richardson (Speaker Recognition), Panayiotis Georgiou, Yannis Stylianou (Speech Analysis); John Hershey (Speech Enhancement); Yifan Gong, Richard C Rose, Tara Sainath, Hagen Soltau, Shinji Watanabe, Kai Yu (Speech Recognition - Acoustic Modeling); Murat Akbacak, Hakan Erdogan, Olivier Siohan, Andreas Stolcke (Speech Recognition – Other); Srinivas Bangalore (Speech Synthesis)
Congratulations to all the new and reelected members, and many thanks to all those who were willing to serve on the Committee and run in the election. We encourage those who were not elected this time to put their names forward again next year!
As always, the elections were the team work of the election subcommittee (with fellow members Larry Heck, Doug Reynolds, Steve Renals, and Moriya Takehiro), and we would like to thank the International Computer Science Institute in Berkeley for hosting the nominations web page. A special thank you is in order to our outgoing SLTC Chair Bhuvana Ramabhadran, who, as every year, devoted a considerable amount of her time to make the elections go smoothly and on time.
