0:00:15i have become a nested into
0:00:18today i will see a presentation on this topic and be compatible dependent you can
0:00:25buy jones for the difficulty
0:00:28i'm cutting question and a domain but that it out in director who have role
0:00:32and gt
0:00:36at fast and
0:00:38i'm trying to show up background to about this study
0:00:44they the gusto disability or
0:00:47mann and thompson or information in it may be because
0:00:52you could be built into the discourse structure as a question is that a tree
0:00:56structure
0:00:58and generate a rst discourse telling about data you below by a strong how to
0:01:04be used in many in can be applications
0:01:07because three thirty two for my and computationally intractable
0:01:13for example i was but have be used for automatic discourse parsing and they some
0:01:21additional autocorrelation
0:01:24and so on
0:01:27on the other hand and we simply some of these chats
0:01:31have proposed another strategy that the discourse given basically
0:01:38they planned event initiate more about the question she
0:01:42in some cases you
0:01:45it's pdas shows
0:01:47that is that it's all text summarization for something they are well-known methods
0:01:54including markers
0:01:56text summarization method
0:01:59a u g
0:02:01at the at data
0:02:05also propose a pity summarization mister based on typically shady coding and
0:02:11he got
0:02:12hi point
0:02:14i don't
0:02:15marcu and other
0:02:18the that's but that
0:02:22dependency based discourse syntax have used a potential to improve the modeling or
0:02:29a wide range it'll take the based applications
0:02:34how it but unfortunately the
0:02:38it is no large scale dtd whole but in english
0:02:43little research it's hard try to create e d compact
0:02:49and i think i you know and the at and propose to on but
0:02:55r is the gtd a two d dct that
0:03:00in this lack a we compare that to combine them
0:03:05i love a seasonal comparing the combined you are mr a as follows
0:03:11after that we make the two missiles algorithm
0:03:15if i to create
0:03:17it can
0:03:19a which format this complex
0:03:22room that depend basically asked would be point of view
0:03:25that which format that and i more accurately by automatic have that
0:03:33finally which comments almost equal to critics summarization
0:03:41i an example
0:03:43rst discuss the
0:03:45you know
0:03:47e one e to it and corresponds to elementary use cost you need in that
0:03:54at
0:03:58any valuing excluded by curly brackets in the same sentence
0:04:05i'm just saying
0:04:07a domain but are going to buy a rhetorical relation and of course
0:04:14identity of discourse unit
0:04:17each discourse unit can see so i don't cool s
0:04:23and i mean
0:04:23and you go yes they reach
0:04:26expresses more important information down
0:04:31on the other hand is
0:04:33means that started i state each added within
0:04:37some fourteen information
0:04:42according variations are also separate it into two types
0:04:47one any more nuclear and you have a multinuclear
0:04:53mallets union has only a multinuclear relation e
0:04:58one and the one s is close to any while a multinuclear relation pinks is
0:05:06that there are no discourse you
0:05:10the recovery shown that can be also glue to use a detailed class q that
0:05:17share it we gotta meaning such as and i've already shown that say nearly thin
0:05:23part of a conditional and so on
0:05:27actually i will not try to home but d r s t e d structure
0:05:33into discourse dependency tree structure
0:05:37also next i really to go into the t k to all their combined
0:05:45well for one you will hear you conditions
0:05:49is
0:05:49that's you know entity
0:05:52methods have at the same combine
0:05:56i just example quality so already more nuclei detection
0:06:05you manipulation each you can see so and i don't and
0:06:12and a because any more important element the s
0:06:17so
0:06:18you know anybody could yes you wanted to
0:06:23in this really show a
0:06:27it belongs to be equal
0:06:30at the right
0:06:34all other education models you don't two d because the guy
0:06:42the next the lead to a she you don't to e
0:06:48and to the next higher data
0:06:52you don't do
0:06:56and finally a
0:06:59according if you don't have to be and to be a bad the
0:07:06to the aperture
0:07:09if
0:07:12and iris we discuss three
0:07:14consider so only multinuclear relations
0:07:18takes you out and v mister produce the same dependency graph
0:07:25how about four one multinuclear relation
0:07:28you know and d
0:07:30a use different way to combat
0:07:36at least example has brown and multinuclear relation
0:07:44you skiing
0:07:47that if the most and he's a
0:07:49mostly in both that
0:07:51quality the most important
0:07:53and the remaining this three d
0:07:56you got
0:08:00in this case the and e
0:08:03a building a you can see the old
0:08:07to see
0:08:19the next either a
0:08:21a middle to see because if we just sat there i
0:08:27and the to the next higher level or a c belongs to be
0:08:33and finally at the attached to the partition
0:08:39i one the other hand
0:08:40easy house cleaning
0:08:43or end of the three d
0:08:45and you want
0:08:47and four or c t and that's to the an excellent very many
0:08:55the next iteration more new variational
0:09:02the next door neighbour manipulation you have a question the
0:09:06it's almost and
0:09:08the us
0:09:10the most in speech and the main a and it could you know to be
0:09:15for all x p long to see
0:09:19a because she the leftmost nucleus
0:09:25in an image which haven't had variation but at that the
0:09:31is higher than in any and or c
0:09:34the anti
0:09:36you are going to e
0:09:39finally need you know the active shuddered
0:09:44now we can see the different either be you know and be sixteen
0:09:50and you can see the most important difference you that you know in a complete
0:09:55you see many crossing act
0:10:01that's you know methods have another serious program
0:10:06by most sentence si
0:10:09for example be strapped rst discourse the
0:10:14but like you know metal all possible with the simple beauty restructure but then test
0:10:21in this example
0:10:23the sentence
0:10:25consisting or used and eighty eight
0:10:28e nineteen
0:10:30had to
0:10:32collected at that same as you we always the student boutiques about tell the
0:10:41to solve this program used by the proposed a post editing missile
0:10:47pitch attitude the
0:10:50you we extend the lexical model the one
0:10:54that is the most
0:10:57you think example that do you need you nine
0:11:02you can be attached
0:11:03two in the next the most into uni you see it
0:11:10it see that very simple updating
0:11:13but as shown in the next that i used simple was it being you see
0:11:20the complexity or that's reality deep
0:11:23format the rest carry
0:11:29now we can have got you there are two methods
0:11:34you may know that we combat its three hundred to eighty five r is the
0:11:40dg in this is cool dependence you
0:11:46last week and it got still
0:11:49the difference is that you max happening in the number of no leaving the text
0:11:55it is well on the whole
0:11:59max patterning need the maximum number four
0:12:04act on that a scroll on i picture due to some belief no
0:12:09not me that
0:12:10number one or leaving that x
0:12:13x
0:12:16roamed easily that it is could that
0:12:19there you know dtd format e
0:12:21what role and channel
0:12:24a fairly format
0:12:28and that we can see that the you for it that's in but it
0:12:34the large gap between be a
0:12:38here we use more complex change goal
0:12:41costing tax
0:12:42e d u
0:12:45you can see that you or weight project e
0:12:49a you have a wide there is no question e
0:12:52he is added
0:12:55why are you know but for not have many cross syntax
0:13:00okay but i'll a proposal was in that are generally use many costing that scroll
0:13:06on the else
0:13:07but
0:13:11x
0:13:12and we get a good use their own automatic discourse dependency parsing
0:13:19we compared to have a
0:13:23one d in this paper that
0:13:25respectively provide you see
0:13:28and on predicting and distinctive from a user
0:13:34that at a charity are
0:13:37and which produce cheese
0:13:38and r is the discuss t for my but can and then be compactly
0:13:44you do you see that each combined you will
0:13:55problem use table we have assumed that we suppose that
0:13:59so any usually you did it directly or it can but it wrong
0:14:06i do you see
0:14:07because to me
0:14:11i have a
0:14:14you got
0:14:18i post us sixty six point
0:14:22and you know got low a
0:14:26fifty five point
0:14:30well of accurately db them up on break down the tree format
0:14:35i have a period of time
0:14:38fe
0:14:39which you do it that we call this is directed at different
0:14:45next we conduct experiments on text summarization
0:14:49to compare the different dependencies that yet
0:14:57be used to
0:14:59you know text summarization network based on
0:15:02for any all i'll a better discourse the game quickly
0:15:07retrieved while the current faded mister
0:15:11it's text summarization moral probability i do you have some program and
0:15:17if you migration can be stored but in figuring out programming
0:15:23for the details pretty that you fight tutees with your paper
0:15:30but at a whole month or text summarization
0:15:34fatty documents
0:15:36which
0:15:39it man made of course online
0:15:43i that
0:15:45but for do d t
0:15:48you know you did you hallmark got it high boring
0:15:53the other formats
0:15:55but the difference is not the last
0:16:00the hunt and when you want a dtd
0:16:06also go high sampling
0:16:08in the different see that
0:16:13well that it indeed the whole paper is there a g e
0:16:19combating from i d e
0:16:21is it because you know dvds over the in is in this paper
0:16:27what religion channel so important element that you really
0:16:34to active channel two and it can be ut extracting back to be mapped to
0:16:39a program
0:16:41i do you see that the wrong at advantage or you know was d v
0:16:44d format
0:16:46or but we did you format
0:16:51in calculating used presentation we need to do so
0:16:56some of the key point
0:16:59used by the compared to combine from
0:17:03r is the dt in the pdtb
0:17:07well you make that i would be if it is clear
0:17:12and the
0:17:15well experimentally that we show that
0:17:19but for mapping what are correct target
0:17:23i'll about for text summarization you know the method has come at the boundaries
0:17:29although mister
0:17:33if you tell that we need you the two digit format score of the task
0:17:39of white sample and discourse
0:17:41three banks by dividing mister doctor adding
0:17:45and the classifier evaluation
0:17:48in you to laugh
0:17:49we did not use
0:17:52maybe a colleague are not there are so each feature
0:17:56so it would you experiments using a particular
0:18:00we plan to investigate the
0:18:03two d for my if is used it
0:18:06adorable my presentation and give a metaphor
0:18:40a offline and which one and two
0:18:46apply
0:18:47including all of these two but
0:19:25no impact know it
0:19:27really
0:19:31something
0:19:55i g
0:19:57that
0:20:02actually
0:20:08i'm going
0:20:30but in a
0:20:33two
0:20:34vol
0:20:38easy to be a high enough that okay you