| 0:00:15 | so first thank you for coming to my presentation |
|---|
| 0:00:18 | in the first part i'm going to introduce a if you a notion of all |
|---|
| 0:00:22 | of the problem i want to tackle |
|---|
| 0:00:24 | in the second part i'm going to show how to model and solve this problem |
|---|
| 0:00:29 | in linguistic terms |
|---|
| 0:00:31 | and finally i'm going to implement a solution |
|---|
| 0:00:34 | in a formal system |
|---|
| 0:00:36 | so here |
|---|
| 0:00:38 | we have a very simple example involving a discourse relation that you didn't come to |
|---|
| 0:00:44 | the party to one out of style so we expected to infer an explanation relation |
|---|
| 0:00:49 | between the two sentences |
|---|
| 0:00:50 | obviously this explanation can be made explicit with a discourse connective |
|---|
| 0:00:55 | jane didn't come to the problem because she one out of style |
|---|
| 0:00:59 | so in english you have three main categories of discourse connectives the first are coordinating |
|---|
| 0:01:05 | conjunctions with and all other than subordinating conjunction with because or even though |
|---|
| 0:01:12 | and obvious instead of the white ball for |
|---|
| 0:01:15 | so one of the phenomena i'm interested in it what we call the none alignment |
|---|
| 0:01:21 | of syntactic and discourse structure |
|---|
| 0:01:23 | so there is an alignment or mismatch |
|---|
| 0:01:26 | when there is no mapping between the syntactic arguments of a discourse connective and the |
|---|
| 0:01:30 | arguments of the discourse relation that is lexical |
|---|
| 0:01:33 | so a very basic example |
|---|
| 0:01:35 | a concern at the bills because they have only one syntactic arguments |
|---|
| 0:01:39 | but you have to discourse arguments for the discourse relation |
|---|
| 0:01:43 | for example blah than jane went to paris |
|---|
| 0:01:47 | and so this |
|---|
| 0:01:50 | for those women under we also consider does not structure all |
|---|
| 0:01:54 | and are often a mobile and of course |
|---|
| 0:01:57 | but i'm interested in other kind of mismatch so to understand that is first look |
|---|
| 0:02:03 | at this discourse |
|---|
| 0:02:05 | jane is very talkative as she told me for example that you had an apple |
|---|
| 0:02:09 | for access to the there is no mismatch here because an example of j being |
|---|
| 0:02:13 | very talkative is telling me about breakfast and so |
|---|
| 0:02:17 | but now |
|---|
| 0:02:18 | i see a trend is are lucky jane told me for example but he won |
|---|
| 0:02:22 | the lottery last week now one is an example of friends being very likely is |
|---|
| 0:02:26 | not change holding anything but that he one of the let's stick to here we |
|---|
| 0:02:29 | have a mismatch and on the line |
|---|
| 0:02:32 | yes in this body blocking the discourse argument respond to a corpus of the syntactic |
|---|
| 0:02:37 | so you have this is actually supporting conjecture |
|---|
| 0:02:42 | so here no mismatch between the j thing that here in europe |
|---|
| 0:02:47 | so them becomes less |
|---|
| 0:02:50 | if |
|---|
| 0:02:51 | i think you indicated here |
|---|
| 0:02:53 | but a fragment time for a fragment of parallel data think that you know which |
|---|
| 0:02:57 | will email |
|---|
| 0:02:59 | this kind of sentences are |
|---|
| 0:03:01 | and the light with usually with a mismatch because we |
|---|
| 0:03:08 | the speaker intends to |
|---|
| 0:03:11 | express the contrast between going to pair and not going to t |
|---|
| 0:03:15 | so those high enough examples are discussed |
|---|
| 0:03:17 | like the intention context |
|---|
| 0:03:19 | no relation to the penn discourse treebank so this kind of example annotated with that |
|---|
| 0:03:24 | such mismatch in the pdtb |
|---|
| 0:03:28 | but it has been question by a hard to support their uniformity hypothesis according to |
|---|
| 0:03:34 | uniformity there is no such mismatch |
|---|
| 0:03:36 | according to you do uniformity the them the contrasting for be would be between found |
|---|
| 0:03:41 | when comparing and for jane think that knowledge |
|---|
| 0:03:45 | but the reasoning in that |
|---|
| 0:03:47 | you need if a and b are in contrast you can say that a and |
|---|
| 0:03:51 | someone saying we are in contrast but the problem in that this kind of preventing |
|---|
| 0:03:56 | the times in to extend to all the kind of a discourse relation |
|---|
| 0:04:01 | if an example of a is b |
|---|
| 0:04:04 | you don't wanna say that an example of a is someone saying be the women |
|---|
| 0:04:08 | going to follow the uniformity purpose of this work considering kind of mismatches |
|---|
| 0:04:14 | you have also very similar in a minute with implicit relations |
|---|
| 0:04:17 | a friend did not come to the party jane told me he was taking care |
|---|
| 0:04:21 | of his gets |
|---|
| 0:04:25 | so when you have such a mismatch |
|---|
| 0:04:28 | you have something and income taxes or according to don't listen rambo in i there |
|---|
| 0:04:35 | is a commitment of the speaker to once the content of the report |
|---|
| 0:04:39 | you agree with one jane told it is what we call the religion affected information |
|---|
| 0:04:46 | but for sensor and after there is no official religion faculty but you expected to |
|---|
| 0:04:52 | infer |
|---|
| 0:04:53 | hedge relations or possible explanation and because explanation is a very the correlation that it |
|---|
| 0:04:58 | may be that it is possible that afraid was taking care six get |
|---|
| 0:05:03 | i so |
|---|
| 0:05:05 | i don't know exactly what is true do we |
|---|
| 0:05:10 | do you like in five when you want to find you necessarily agree with what |
|---|
| 0:05:15 | jane told you or do you |
|---|
| 0:05:19 | at least come to their it's possible i don't know exactly so we're going to |
|---|
| 0:05:23 | talk about that like |
|---|
| 0:05:25 | and see how we can almost constant |
|---|
| 0:05:28 | bit of technology so those mismatches in the whole action verbs like to be used |
|---|
| 0:05:34 | to know all those of people would like to say and tell a these four |
|---|
| 0:05:40 | and with distinguishing between the parenthetical user but it's okay one predicate is not possible |
|---|
| 0:05:46 | of the discourse argument so when there is an on the alignment and on the |
|---|
| 0:05:50 | other hand the and attentional use of a t v so when the predicate is |
|---|
| 0:05:55 | included in the discourse segments so when there is no that the line |
|---|
| 0:06:00 | yes very important not all not all connectives are compatible with a session of alignment |
|---|
| 0:06:07 | for example we are cool lexicalising an explanation you can say that did not come |
|---|
| 0:06:13 | to the party because he was taking care of its good at you cannot instead |
|---|
| 0:06:17 | an evidential weight you cannot say predicting outcomes of partly because jane says he was |
|---|
| 0:06:21 | taken care |
|---|
| 0:06:22 | and the problem comes from the connective because you can do that with an implicit |
|---|
| 0:06:27 | relation you can say for a given that comes partly j said he was taking |
|---|
| 0:06:31 | care of is good |
|---|
| 0:06:32 | the problem comes from the connective blocking their the infinite shown in fig |
|---|
| 0:06:39 | so our we're going to mortars so as to the to kind of other we |
|---|
| 0:06:44 | also is that are distinguished by a high amount so on the one hand you |
|---|
| 0:06:51 | have sensible a direct orders so they show good syntactic integration tool a measure close |
|---|
| 0:06:55 | and they provide some time for local information on the intensity in the matrix |
|---|
| 0:07:02 | for example jane left of the written because you have an appointment this is a |
|---|
| 0:07:07 | sensible close according to like them |
|---|
| 0:07:10 | on the other hand you have paris for and the realtors as they are much |
|---|
| 0:07:13 | less into were integrated to the matrix level and that's function is more to structure |
|---|
| 0:07:19 | the discourse so they can express concession or some but one relation for interpretation of |
|---|
| 0:07:25 | the matrix roles for example in eight |
|---|
| 0:07:28 | fragile estimates for less the rest and even though here not finished eating so you've |
|---|
| 0:07:33 | got some the concession this is a powerful and real close to record |
|---|
| 0:07:39 | so she wants many differences in syntax-semantics probability and so on between those two kind |
|---|
| 0:07:45 | of data is that what we see that another different it was performed so that |
|---|
| 0:07:50 | are compatible with last year's while sensible one or not |
|---|
| 0:07:54 | so we already thing that in the previous example but |
|---|
| 0:07:58 | please consider you know and b goes |
|---|
| 0:08:01 | so you can you can have |
|---|
| 0:08:03 | the intangible like in nine so no mismatch for the case the policy |
|---|
| 0:08:09 | even though jane said he'd and should be one thing that you can have a |
|---|
| 0:08:13 | mismatch like in we fight change the party even though j said you hadn't finished |
|---|
| 0:08:17 | writing his research couple |
|---|
| 0:08:19 | but you can and do that we because so you can have to be like |
|---|
| 0:08:24 | in ten eight contains the partly because jane said she wanted to talk to him |
|---|
| 0:08:29 | but you cannot have any and then eventually likely to be fed into the partly |
|---|
| 0:08:34 | because james as he finished writing is research cable |
|---|
| 0:08:41 | so how could one expenses so well stream no i tell you audio that are |
|---|
| 0:08:46 | middle connective a considered and of course not to what i'm going to say that |
|---|
| 0:08:51 | in fact even conjunctions are not for the structure of the kind of selection mechanism |
|---|
| 0:08:55 | able syntax but with more the that i'm going to model as an unable to |
|---|
| 0:08:59 | the y about syntax |
|---|
| 0:09:02 | we couldn't better use examples of all useful so a connective sse |
|---|
| 0:09:07 | and so i don't want to postulate any syntactic n b greedy in this kind |
|---|
| 0:09:13 | of sentence that sentences so that the if there is a mismatch or not |
|---|
| 0:09:18 | the this syntactic structures same but there is this selection mechanism involved in with the |
|---|
| 0:09:24 | connective that will explain the different kind of interpretation |
|---|
| 0:09:29 | yes and obviously this selection mechanism always different constraints force and for every from all |
|---|
| 0:09:33 | sentences |
|---|
| 0:09:34 | those can find a |
|---|
| 0:09:36 | first |
|---|
| 0:09:37 | a discourse arguments must have been introduced by the corresponding syntactic argument but you cannot |
|---|
| 0:09:42 | select |
|---|
| 0:09:43 | anything like when you're discourse connective you need to select something that is related to |
|---|
| 0:09:47 | you that syntactic arguments and a central connective cannot they compose a closed headed by |
|---|
| 0:09:53 | a navy so that you to access the report but to parry for one can |
|---|
| 0:09:58 | so with us to a constraint you can really account for the data we would |
|---|
| 0:10:04 | seen that before |
|---|
| 0:10:05 | but a in addition are considered as a something that is related to sequences of |
|---|
| 0:10:10 | conjunctions |
|---|
| 0:10:11 | a congestion one b conjunction juicy but i'm not going to dwell |
|---|
| 0:10:18 | yes so i'm going to implement that it is now i mean in a formal |
|---|
| 0:10:23 | system called a continuation semantics so it's based on lambda calculus it has been used |
|---|
| 0:10:29 | to model the syntax-semantics interface and |
|---|
| 0:10:33 | also some aspects of the semantics pragmatics interface and that's the kind of it we're |
|---|
| 0:10:37 | going to use right now |
|---|
| 0:10:39 | so and you are then used to model f one for nominal for |
|---|
| 0:10:46 | so in well you got a very standard high priced for proper noun like jane |
|---|
| 0:10:51 | is a function with an argument pete's and individual predicate and j just take that |
|---|
| 0:10:57 | these two of the consulting |
|---|
| 0:10:59 | now for context sensitive |
|---|
| 0:11:02 | so like of the pronoun she |
|---|
| 0:11:04 | you need some expected the context arguments |
|---|
| 0:11:09 | and now p is not true of any specific constant a viable it's true and |
|---|
| 0:11:14 | individual that has to be retrieved from the context |
|---|
| 0:11:18 | you think what people selection function |
|---|
| 0:11:21 | and in this case the output context use the same every |
|---|
| 0:11:25 | for a predefined expressions such as cats |
|---|
| 0:11:30 | what happened these we introduced by a factor the cat is to this |
|---|
| 0:11:34 | and we aren't information that x is a cat in the context so it can |
|---|
| 0:11:38 | be this information is from might be relevant for subsequent enough |
|---|
| 0:11:43 | i don't and the context c |
|---|
| 0:11:46 | is transmitted from sentence to sentence for this the discourse of the operator |
|---|
| 0:11:51 | a very simple example she's with things that we wish additionally get something like that |
|---|
| 0:11:57 | well |
|---|
| 0:11:58 | it's quite funny or just |
|---|
| 0:12:01 | the individual corresponding to she in the complexity |
|---|
| 0:12:04 | i think on the whole this proposition |
|---|
| 0:12:11 | so we're going to we use this model of anaphoric expressions of to model the |
|---|
| 0:12:15 | behaviours discourse connectives to do that's it we're going to add the basic type woman |
|---|
| 0:12:21 | from the for propositional reference so that matters the different from the g the choose |
|---|
| 0:12:26 | volume type |
|---|
| 0:12:28 | because these markers are going to be the arguments of discourse relation |
|---|
| 0:12:33 | so i consider any sentence describe a propositional market that is given as argument but |
|---|
| 0:12:39 | can also introduce watermark as for example when there is a number that roles |
|---|
| 0:12:45 | and so now he's we think is something like that you get this condition e |
|---|
| 0:12:49 | liable all the proposition law and what happens is based at that |
|---|
| 0:12:54 | i e is a proposition that use this thing and |
|---|
| 0:13:00 | and of course because propositions control we need a true a pretty eight that just |
|---|
| 0:13:05 | a tad bit hes an argument denotes a superposition |
|---|
| 0:13:10 | is somehow the discourse update it looks like that so what how do you continue |
|---|
| 0:13:16 | to discourse d with the new sentence s |
|---|
| 0:13:19 | so what's encountered here but we introduce a new proposition along with played that this |
|---|
| 0:13:25 | proposition one denotes the true population and that without recognition of s |
|---|
| 0:13:30 | given this the potential market |
|---|
| 0:13:33 | for and maybe such a thing |
|---|
| 0:13:35 | what is important in the we introduce a positional marker you problem that for the |
|---|
| 0:13:39 | embedded everything that |
|---|
| 0:13:42 | e interpretation that s |
|---|
| 0:13:44 | things in prior and we start executing the and the embedded close given prime |
|---|
| 0:13:49 | we don't like that in practice true because |
|---|
| 0:13:53 | i think it does not give them so someone can think something that |
|---|
| 0:13:57 | now the last time |
|---|
| 0:13:59 | in the this pretty big channel for the because |
|---|
| 0:14:05 | so what's importantly we have to augment a and b for the two |
|---|
| 0:14:10 | very first oppositional a quite a that is introduced and passed to the recognition and |
|---|
| 0:14:16 | then |
|---|
| 0:14:17 | a propositional a kiwi is introduced and passed to the application of me and then |
|---|
| 0:14:21 | the main property should he is described as an explanation holding between to do things |
|---|
| 0:14:27 | to augment that has to be selected |
|---|
| 0:14:30 | and so you got for instance this term select an appropriate propositional markets in the |
|---|
| 0:14:36 | context the system |
|---|
| 0:14:37 | that is really do not have to satisfy the three constrain the this section six |
|---|
| 0:14:42 | and seven see for a central |
|---|
| 0:14:46 | a connective so they it needs to satisfy the this we construct a for a |
|---|
| 0:14:53 | note that connect points and although it's gonna be very similar |
|---|
| 0:14:57 | the only difference |
|---|
| 0:14:58 | is going to be the lexicalized discourse relation and the selection function is gonna be |
|---|
| 0:15:03 | this cell speech section function |
|---|
| 0:15:05 | that is satisfied only to constrain |
|---|
| 0:15:09 | so if we consider the sentence print came to the party although jane things will |
|---|
| 0:15:16 | think what we get it this a big thing |
|---|
| 0:15:20 | so what happens is we expect that for k and this is okay |
|---|
| 0:15:25 | then |
|---|
| 0:15:26 | again |
|---|
| 0:15:27 | in trying to be a prime |
|---|
| 0:15:30 | he is think |
|---|
| 0:15:32 | and we have a conversation between holding between those two times |
|---|
| 0:15:35 | and in this particular case there is only one |
|---|
| 0:15:40 | a positional marker related to a in this context so the first argument is going |
|---|
| 0:15:45 | to be necessarily the case for making the possible |
|---|
| 0:15:49 | but for this argument |
|---|
| 0:15:51 | there is actually to a different position long as related to me satisfying the three |
|---|
| 0:15:56 | that the constraint you have |
|---|
| 0:15:58 | maybe so anything that was sick and it prior |
|---|
| 0:16:02 | is applied |
|---|
| 0:16:03 | which is only a few was so it is but two possibilities are available and |
|---|
| 0:16:10 | based on the world knowledge this lp algorithm will be able to decide that the |
|---|
| 0:16:16 | correct one is just a in primes |
|---|
| 0:16:20 | not all there is less than that with since like you so french state on |
|---|
| 0:16:25 | because j if you will see |
|---|
| 0:16:28 | in this case |
|---|
| 0:16:30 | instead of lp you will have to see so about satisfying different constraint and the |
|---|
| 0:16:37 | only possibility for those of an argument you think you primes a little or jane |
|---|
| 0:16:41 | think you will see and in this case it it's and natural interpretation |
|---|
| 0:16:46 | so we can explain |
|---|
| 0:16:49 | one |
|---|
| 0:16:49 | this is correct and this is not |
|---|
| 0:16:53 | yes |
|---|
| 0:16:54 | us a lot about this mapping consequences that i mentioned so if we think they |
|---|
| 0:16:59 | consider a some very elite the actual so before and there exist the true confession |
|---|
| 0:17:05 | holding between two proposition then the to proposition must be true |
|---|
| 0:17:10 | if we do that's we get just from where jennifer actually information |
|---|
| 0:17:15 | this by dollars in mimo |
|---|
| 0:17:18 | but if we decide to slightly change that sounds for the connective so that they |
|---|
| 0:17:23 | introduced hedge relations when then selecting the structural argument |
|---|
| 0:17:27 | we get the interpretation proposed by hands and eyes |
|---|
| 0:17:33 | so what have been done so we found that the distinction between a parenthetical and |
|---|
| 0:17:39 | parenthetical you deserve at this is explained in terms of discourse unfolds and that the |
|---|
| 0:17:45 | distinction between sends one powerful conjunction include specific argument selection constraints of those is thoughtful |
|---|
| 0:17:51 | and we show that |
|---|
| 0:17:54 | the average height or discourse connectives can be implemented compositionally using continuation |
|---|
| 0:18:01 | thank you for your attention and please ask me any questions |
|---|
| 0:18:35 | so we have to mention that this is very theoretical work so i'm not trying |
|---|
| 0:18:41 | to actually implement the system that is going to do that because it relies on |
|---|
| 0:18:47 | like good syntactic analysis and it's very hard to get |
|---|
| 0:18:52 | and plus if you want to analyze the full discourse you need to have like |
|---|
| 0:18:56 | to affect syntactic analyses for all the sentences so that the and so it's |
|---|
| 0:19:01 | not |
|---|
| 0:19:02 | two days not practically doable but its aim of this work was to explain the |
|---|
| 0:19:08 | why some connectives a compatible with mismatches wine some not like more linguistic or not |
|---|
| 0:19:33 | well |
|---|
| 0:19:34 | so you can is selection function so you can plug any kind of algorithm you |
|---|
| 0:19:39 | want so probably something related to get the resulting system if you're improve the that's |
|---|
| 0:19:46 | use |
|---|
| 0:19:47 | so the competition like the conditional computational aspects are going to be related to what |
|---|
| 0:19:53 | kind of specific function you implement for this we can do you gotta you could |
|---|
| 0:19:57 | have a very basic heuristic or input other much texas |
|---|
| 0:20:30 | yes |
|---|
| 0:20:58 | so the thing is when |
|---|
| 0:21:01 | when you like the way it's and this kind of us a discourse on a |
|---|
| 0:21:06 | light in the literature and i will not have once you infer the explanation relation |
|---|
| 0:21:12 | you're kind of committing to the actual is a possibility of the other two something |
|---|
| 0:21:16 | in particular |
|---|
| 0:21:37 | so |
|---|
| 0:21:40 | but in this case you are not going to infer the explanation relation or you're |
|---|
| 0:21:43 | going to attribute this explanation to jane not to use all |
|---|
| 0:21:57 | actually it's less obvious when you have implicit discourse relation we could you can be |
|---|
| 0:22:01 | page could be attributed someone else but if we go back to example with for |
|---|
| 0:22:06 | example like a friend is lucky jane told me for example that he wanted the |
|---|
| 0:22:12 | last we like we last week it seems very hard to say that she |
|---|
| 0:22:17 | not an example was lying or so in this case you will kind of committing |
|---|
| 0:23:01 | i know if you use this kind of children you've prepositional phrases like a according |
|---|
| 0:23:06 | to j so they behave differently but they are really syntactically parenthetical and so you |
|---|
| 0:23:11 | can always i whose for any within a connective |
|---|
| 0:23:25 | yes |
|---|
| 0:23:29 | but there is there something we're going on when used according to jane because sometimes |
|---|
| 0:23:35 | it seems like the relation is actually needed to a the like you it the |
|---|
| 0:23:39 | whole the input the edge between phrase is coping of a discourse connective but sometimes |
|---|
| 0:23:44 | not it's |
|---|
| 0:23:46 | a bit more complex |
|---|
| 0:23:54 | i've been here sorry |
|---|
| 0:24:00 | and |
|---|
| 0:24:01 | no well |
|---|
| 0:24:04 | i know exactly like |
|---|
| 0:24:06 | what would be according to me difference is like the scoping relations |
|---|
| 0:24:12 | how it's going to scope of the discourse relation |
|---|
| 0:24:17 | which is |
|---|
| 0:24:18 | both because is syntactically complete very different so you expect that the scoping relations are |
|---|
| 0:24:24 | going to be different and that's probably what |
|---|