0:00:15re
0:00:17okay like to welcome everybody see this
0:00:20a special session a natural language generation for dialogue systems
0:00:24i'm just gonna give a five minute overview of the session then we'll have a
0:00:30couple of long talks accomplish very short talks and then we'll have a panel
0:00:34and
0:00:35about a lot of this kind of came together to organize this so the other
0:00:39organisers of their temper like a kind reader david how corrupt
0:00:43showing or read and very in a research
0:00:47so the
0:00:48what the organizers task is to do is to say like why we would have
0:00:52a special
0:00:54session on a nlg for dialogue systems
0:00:58and you know because you might think well this
0:01:01areas been around for a long time indeed some of the earliest work
0:01:04on computational models for natural language generation was work done in the context of a
0:01:09dialogue system or question answering system like kathy make humans
0:01:13early work for phil collins early work
0:01:17there's also a lot of earlier much earlier work you note going back almost twenty
0:01:22years
0:01:24people doing starting to do statistical
0:01:27work on statistical natural language
0:01:30generation
0:01:31starting with the kind of seminal work of like alien like who showed that you
0:01:35could kind of have a very loosely configured hybrid
0:01:39linguistic statistical representation where you could overgenerate and then you could learn
0:01:44rightly rules and filter out
0:01:47this filter out the data's you could produce and so that works as old as
0:01:51nineteen ninety eight so you might
0:01:52still say like why would you have the special session now and also in the
0:01:57context of darpa communicator conversational dialogue systems
0:02:01i edited a special issue with computers
0:02:04speech and language are not sure language generation for spoken dialogue
0:02:08and the number of people that you know had papers in the special issue including
0:02:12am and i lasso and ridge alex rudnicky mari ostendorf stephanie seneff so a lot
0:02:18of long time people been working on conversational dialogue systems
0:02:22for years
0:02:24but the reason that we wanted to have this special session despite the fact that
0:02:27these a lot of that
0:02:28its of generation for dialogue have been around a long time is that there's been
0:02:33a recent kind of resurgence of interest in a natural language generation because of kind
0:02:40of the
0:02:41ai renaissance i guess we should say all the interest in chat bots and all
0:02:46the consumer products that are out there now like collects and google system
0:02:50the other thing is that the availability of large online corpora like open subtitles or
0:02:57twitter or
0:02:59i you know corpora like that have led have may people wonder whether they could
0:03:03actually use a purely
0:03:04statistical kind of and machine translation approach to produce dialogue turns in an open-domain way
0:03:11so there's been a lot of activity in that area for the last five years
0:03:16and
0:03:18so
0:03:19so there's a lot you know seems to be a lot of different stuff going
0:03:21on of this field and one of the reasons why i wanted to organize the
0:03:25special session was the kind of
0:03:27be able to look at what we can do now with different generation techniques into
0:03:31bring you know
0:03:33it especially in the panel to bring in a panel of experts people who worked
0:03:37on one language generation for dialogue systems and try to get some different perspectives
0:03:42on from
0:03:43from their from their points of view what kinds of techniques which ones don't work
0:03:47which things the ready for primetime it could go into consumer products in which things
0:03:51are still just
0:03:52kind of basic
0:03:55research ideas
0:03:57and when am i am particular interest and i think that of many of the
0:04:00other people to organize the panel who have put out these other challenges but e
0:04:05two e challenge is also the web nlg challenge now is in interest and stylistic
0:04:11variation in some of the classic things the natural language generation sabine able to do
0:04:16in their sentence planner and like and a kind of interest in whether
0:04:21and to and framework is actually without a lot of extra architectural details that kind
0:04:27of model the traditional natural language
0:04:30architecture whether they're actually gonna be able to produce different kinds of
0:04:34stylistic variation like the
0:04:37previous generation of statistical language generations could do
0:04:41so that's kind of why were here
0:04:44and
0:04:46we have
0:04:47too long papers
0:04:49redundancy localisation for the conversation of unstructured responses and that a neural language generation in
0:04:55dialogue paper
0:04:56we have to short papers that will be presented in five minutes the for the
0:05:01panel
0:05:02and they'll be in the poster session later i wanted to have them in
0:05:06in the discussion in our in our minds before we start doing the panel
0:05:11because i think
0:05:13there's a really interesting thing here of a new generation challenge that aims to be
0:05:17a little bit more complicated than what people then you seen in the neural generation
0:05:21framework
0:05:21and then somebody was really
0:05:24hot of the starting box
0:05:27and that
0:05:28the corpus was released in two weeks later they had a they this
0:05:33character the character out of the box model that uses the corpus of this is
0:05:37all kind of very much
0:05:39breaking news
0:05:40and
0:05:42so we can go ahead and
0:05:44get started for the for the main papers and then we'll have the panel at
0:05:48the
0:05:49at the end
0:05:50okay so sebastian