0:00:14 | and thirty one |
---|
0:00:16 | i |
---|
0:00:16 | i'm happy lasttime applied is due to indicate that at university |
---|
0:00:21 | and |
---|
0:00:22 | and today representing a study on are self disclosure in conversation dialogue system |
---|
0:00:31 | so i haven't recovers also if you can't understand i'm happy i said |
---|
0:00:40 | so this was the study was done as part of the cmu my is the |
---|
0:00:44 | i v and is not exactly in two thousand seven |
---|
0:00:50 | well as in just standing there is close to |
---|
0:00:54 | so it's because human conversations of in humans and y are many solutions to think |
---|
0:01:00 | is right here try to achieve the solution and restart it can be in of |
---|
0:01:05 | all those that so propositional function based |
---|
0:01:08 | i have any information of the conversation those that so interactional con functions with just |
---|
0:01:14 | a system like the quantization for one |
---|
0:01:18 | and so in cost functions which is |
---|
0:01:22 | trying to build that of the essential between those who participated |
---|
0:01:26 | so set motion is one of the key social strategy employed in conversation and intimacy |
---|
0:01:32 | between pocketsphinx and interleaved idea of the conversation |
---|
0:01:38 | so many definitions on the one wants to self disclosure |
---|
0:01:42 | i o one is that strong or mean and it in nineteen seventy three which |
---|
0:01:48 | defined as the one a static |
---|
0:01:50 | all opinions or but experiences references by you and wasn't it |
---|
0:01:57 | so |
---|
0:02:02 | so |
---|
0:02:03 | a no side distortion is a very interesting phenomenon in between very well studied by |
---|
0:02:09 | the psychology community |
---|
0:02:11 | in particular because of |
---|
0:02:14 | it's actually the ability to use reciprocity and dyadic interaction |
---|
0:02:19 | so that you must leave the phenomenon vibrates when one participated in a conversation self |
---|
0:02:24 | disclosure |
---|
0:02:25 | the other participants in most competitive set of discourse in this form |
---|
0:02:29 | and there are many explanations on this but the exact called also based on the |
---|
0:02:34 | screen on that's not what is the one hypothesis is that it's a formal solution |
---|
0:02:39 | extreme where the party receiving side distortions this feels obligated but also said this goal |
---|
0:02:46 | i don't know what concerns is that it's a solution conversational where if a |
---|
0:02:51 | it doesn't sound is close and return they feel uncomfortable |
---|
0:02:55 | and i and hypothesis is about social just attraction |
---|
0:03:00 | ones that is close to people and that's close to that of because i |
---|
0:03:05 | as i do not trust in like a by the exact it's not what a |
---|
0:03:12 | distance between a pretty well established a |
---|
0:03:15 | and right of self disclosure and have been reproduced in many studies and shall |
---|
0:03:21 | to be a very strong |
---|
0:03:25 | so subsequent studies also show other ask exercise caution like |
---|
0:03:30 | a self disclosure reciprocity characterizes initial social interaction because people set is closely to try |
---|
0:03:36 | to corpsman model |
---|
0:03:38 | i interestingly you look at a distance to be high school |
---|
0:03:42 | and actually to be better eliciting self disclosure |
---|
0:03:46 | also is not it your relationship between this process and it |
---|
0:03:51 | so it's not expected that a higher amount of self disclosure that make someone like |
---|
0:03:56 | you |
---|
0:03:58 | so |
---|
0:03:59 | that's has been studied for a conversation between human |
---|
0:04:04 | but really interested to know if the same it's that is close to have the |
---|
0:04:09 | thing i think in human machine i work |
---|
0:04:12 | and if it does |
---|
0:04:13 | that would have |
---|
0:04:15 | implications for systems which came to elicit information from the user point o to maybe |
---|
0:04:22 | also more pleasant nice experiences all the methods you task completion |
---|
0:04:28 | but the key axes |
---|
0:04:29 | two months that machines don't have sort since you things of that one so any |
---|
0:04:35 | set this can also apply machine guns that is coming across as dishonest |
---|
0:04:41 | by the starting point is that night maybe force that humans actually something you computers |
---|
0:04:49 | a solution that so almost macabre thinking about n i the second source you to |
---|
0:04:55 | stick study was split |
---|
0:04:56 | in human conversation the human machine and or |
---|
0:05:00 | okay so i'm |
---|
0:05:04 | we are talking now about the context in which we can you contacted is that |
---|
0:05:10 | so everyone who works in dialogue was how difficult it is together data |
---|
0:05:15 | but |
---|
0:05:17 | exactly in two thousand seven amazon had an example right channel |
---|
0:05:22 | we had a noun in some university students to was channel one on amazon devices |
---|
0:05:30 | and so this was a pretty one because |
---|
0:05:36 | we actually get that uses the real world instead of |
---|
0:05:40 | you know i think something expensive test it |
---|
0:05:43 | so you want one of sixteen dialogue instance that hosted on the alex at the |
---|
0:05:48 | right |
---|
0:05:49 | at this could be able to use the united states of the command that's chat |
---|
0:05:54 | and you that's what i didn't the data and it is seen that i don't |
---|
0:05:58 | issues so they didn't know which of the dialog state tracking with |
---|
0:06:03 | so |
---|
0:06:05 | it looks like interesting because |
---|
0:06:07 | i think it was that's what she to end the conversation at any time so |
---|
0:06:11 | a data and happens |
---|
0:06:13 | you domain so please specify started goals |
---|
0:06:16 | so the only reason that even continue the conversation was for that when entertainment |
---|
0:06:21 | so at the end of the conversation user is allowed only at the interaction the |
---|
0:06:26 | scale of one point five based on the that they would interact with the social |
---|
0:06:30 | want to get it also three shows more anybody |
---|
0:06:34 | so three hundred and nineteen out of the fifteen or one thousand five hundred users |
---|
0:06:38 | decided to anybody |
---|
0:06:44 | and well known talking about the dialogue agent that we had in the next upgrade |
---|
0:06:51 | so that i don't agent was based on a finite state machine architecture and what |
---|
0:06:56 | this means that every step was a of the finite state machine |
---|
0:07:00 | with essentially the response that we wanted to give as a dialogue system |
---|
0:07:04 | and transitions what condition don't use the sentiment |
---|
0:07:08 | so just an example of how this might look |
---|
0:07:11 | so madness aside dialogue it and it's the entire house and green |
---|
0:07:16 | in the users say something more stable be reports to like not but not by |
---|
0:07:20 | then minus encyclopedia anything special happen if response posted but in the user to say |
---|
0:07:26 | something that sounds naked |
---|
0:07:28 | in tries to get a sympathetic response adidas them what's wrong so in this way |
---|
0:07:36 | we had essentially the order of topics of the conversation |
---|
0:07:41 | so maybe firstly the user then |
---|
0:07:45 | acknowledge the positive or negative response then you try to talk about the initial |
---|
0:07:49 | where we asked if they are interested in one of the latest tv shows if |
---|
0:07:53 | they said that they are not interested that we asked them about a nice in |
---|
0:07:57 | the if they say they were not interest in the movie i don't want game |
---|
0:08:01 | in this example intuition that we show the movie we did spend some time chatting |
---|
0:08:05 | about that it is that it is that you see you with the |
---|
0:08:09 | then eventually all users what invited to leon you believe that out just choose not |
---|
0:08:14 | appeared again and again what kind of one of these long winded word idioms and |
---|
0:08:18 | they could also choose to exhibit limit in there |
---|
0:08:22 | now a deterministic which was right |
---|
0:08:25 | i don't need to impose taking initiative but al |
---|
0:08:29 | all from the stage one with a conversation initiator shifted to the user |
---|
0:08:34 | and they could talk to a dialog agent about anything and we try to get |
---|
0:08:37 | responses from |
---|
0:08:38 | other sources from the way |
---|
0:08:41 | so on is |
---|
0:08:44 | about i dialogue into |
---|
0:08:47 | but the specifying chat bots that we used in this study would integer different so |
---|
0:08:52 | we randomly assigned the users who interacted with the system |
---|
0:08:56 | one of two chat bots or one of them i think in c is a |
---|
0:09:01 | very high self disclosing chat board from the beginning of the conversation |
---|
0:09:05 | so in that uses when the machine that's how's it going in the user sees |
---|
0:09:10 | not i'm not the channel response with the story about itself with this kind of |
---|
0:09:15 | your because |
---|
0:09:17 | that's it is directly related to there's been chilling a to b and catching up |
---|
0:09:21 | with my friends they just got them and whatever you an expression that they shouldn't |
---|
0:09:24 | one |
---|
0:09:25 | and the machine i mean of the dialog agent does not really had a frame |
---|
0:09:30 | is often to work about it |
---|
0:09:32 | but |
---|
0:09:33 | then the humans experiment to see a play today i quite enjoyed |
---|
0:09:37 | by a group of humans |
---|
0:09:39 | the see a dialog agent that did not end of story about itself it is |
---|
0:09:44 | c is all that's create anything a specialization towards technology and is the next question |
---|
0:09:52 | so now on this was the setting under which we |
---|
0:09:58 | and i'm not experiment |
---|
0:10:00 | but not only interested in identifying when you with the we use i with exactly |
---|
0:10:05 | that are dialog data |
---|
0:10:07 | actually said is close and stick with defining all what we consider the self disclosure |
---|
0:10:13 | so in the context of conversations with a dialog agent we said that self disclosure |
---|
0:10:18 | has to be wanted at and it should be information di otherwise the dialogue agent |
---|
0:10:23 | should market |
---|
0:10:25 | so this does not include non-systematic questions for example so in the example given |
---|
0:10:33 | so if you see when a system see what we collected anything special |
---|
0:10:38 | and the use this is nothing that changes going on physically with my notes today |
---|
0:10:41 | so that what constitutes a disclosure |
---|
0:10:44 | but what exactly fifty times in the movie a time to learn the user sees |
---|
0:10:49 | i need identity of a bit bigger block on dataset disclosure because it is adjusted |
---|
0:10:53 | it it's possible question and that you have any extra information |
---|
0:11:00 | so in this paper a three hundred and nineteen conversations were labeled for a user |
---|
0:11:07 | self disclosure |
---|
0:11:08 | and we manage to get a substantial agreement on there |
---|
0:11:14 | by |
---|
0:11:16 | we actually had a much larger corpus of conversations which it was not possible |
---|
0:11:21 | human annotators to allocate every user utterance for self disclosure |
---|
0:11:26 | so what we do this we built an svm classifier |
---|
0:11:29 | are trained on this corpus to be able to the entire corpus for all occurrences |
---|
0:11:34 | of the service goes |
---|
0:11:36 | so this is the things just is designed to say one going to justify |
---|
0:11:40 | but the classifier or a accuracy of ninety one point seven percent and or |
---|
0:11:47 | f one score of sixty seven percent so it was fairly a little bit |
---|
0:11:53 | okay so now we had |
---|
0:11:56 | the user utterances which |
---|
0:11:59 | one instances of self disclosure because we got a classifier to label the whole corpus |
---|
0:12:04 | i mean or when a machine is close because we designed this is so now |
---|
0:12:09 | that's allowed us to study the effects of status close to |
---|
0:12:12 | so are the first one if you want to start with lexical |
---|
0:12:17 | so |
---|
0:12:19 | what we did this we studied |
---|
0:12:21 | in how many users are disclosed and done before to sell disclosing user utterance |
---|
0:12:27 | is the of the machines at school and we note that users was significantly more |
---|
0:12:32 | likely to set the schools following of machine that the solution |
---|
0:12:36 | then when a machine didn't set is close |
---|
0:12:39 | you know at a door immediately following the for instance of machine self disclosure |
---|
0:12:47 | and we found its users were much more likely to set it is close even |
---|
0:12:52 | at the beginning of the quantization of the o |
---|
0:12:54 | matching after a system set is close |
---|
0:12:59 | so we set our study other question |
---|
0:13:02 | the project and a conversations but initial user self disclosure actually longer |
---|
0:13:08 | an internal yes there were a they were significantly lower |
---|
0:13:14 | we also want to study if you that's what is close initially do they disclose |
---|
0:13:19 | of the conversation and i don't know that if someone doesn't so close initially |
---|
0:13:25 | much less likely to sell it is close to the correlation |
---|
0:13:30 | so |
---|
0:13:32 | i don't know question we started was a user will choose not to sell disclose |
---|
0:13:36 | initially estimated as in just employing machine interest and this is kinda based on the |
---|
0:13:42 | motion of people having no god it was not |
---|
0:13:46 | so the way we tested this was a bit a reference to i would do |
---|
0:13:52 | so we are able to set was initially one more likely to fail to one |
---|
0:13:57 | and only if they do play the work in for how long deeply working |
---|
0:14:03 | so yes a the users who a source not a set of is close initially |
---|
0:14:07 | what actually much less likely to pay to what they also if needed in the |
---|
0:14:11 | volume deflated what a much shorter time than |
---|
0:14:14 | a user's we chose to say it's close |
---|
0:14:20 | next we test the effect is close right in the likability in human |
---|
0:14:25 | machine interaction |
---|
0:14:27 | so since avalon provided us with sleeping of well with the people for a given |
---|
0:14:32 | by the system again we use it as a proxy for online okay |
---|
0:14:36 | and we thought we had a conversations make you would sell disclosed a lot i |
---|
0:14:41 | mean the delayed of what vector |
---|
0:14:43 | and |
---|
0:14:44 | the and that is we don't know |
---|
0:14:48 | so we wouldn't really find any correlation between over the user ratings and the a |
---|
0:14:53 | mole this to disclose in the conversation then and we couldn't find a difference in |
---|
0:14:58 | the ratings of the statistic located at what's as both within |
---|
0:15:02 | and he couldn't find a difference in the ratings for conversations which have high service |
---|
0:15:06 | also insist on with patients with discourse |
---|
0:15:12 | and so it and a few movies |
---|
0:15:16 | by we study the effect of set is close in a real-time that skin spoken |
---|
0:15:22 | dialogue system give model being users in the real world of amazon alex that thanks |
---|
0:15:28 | to a amazon |
---|
0:15:30 | and what we found is that indicators of reciprocity have been even in human machine |
---|
0:15:35 | quantization |
---|
0:15:36 | and that |
---|
0:15:38 | by the way how with are authentic as close as efficiently we can characterize the |
---|
0:15:43 | behavior while the quantization |
---|
0:15:45 | well what we also identified anything at relationship between san exclusion and like |
---|
0:15:51 | thank you |
---|
0:15:58 | first |
---|
0:17:04 | the question so it's jules so well they're not aspects to self disclosure to adapt |
---|
0:17:11 | of self disclosure elements of self disclosure and the study considers the only by at |
---|
0:17:16 | o |
---|
0:17:19 | future work would be to do this i'll probably but to do both positive and |
---|
0:17:26 | negative as well or two of the data but the on sentence that i don't |
---|
0:17:32 | more better we would find or relationship between self disclosure and liking |
---|
0:17:37 | because i even in psychology it's not remote that |
---|
0:17:42 | so that was the study in nineteen seventy three i think about what cost |
---|
0:17:47 | would say that |
---|
0:17:49 | so he divide itself is closer to three categories do medium and high essentially |
---|
0:17:54 | and he found that and one set is closer to result in collecting but i |
---|
0:17:59 | said distortion actually resulted in this and i think |
---|
0:18:02 | and the reason is that i don't trust |
---|
0:18:04 | people who say disclose very high |
---|
0:18:09 | so that maybe |
---|
0:18:10 | and interesting finding |
---|
0:18:16 | was |
---|
0:18:29 | i even like you an extra a i think that someone we gotta because they |
---|
0:18:38 | have a block so well |
---|
0:18:41 | i don't know anything that alex is hopelessly but an x and you might be |
---|
0:18:45 | ignored |
---|
0:18:47 | but they are this also five shows that people will be back to what we |
---|
0:18:54 | so that also are similar |
---|
0:18:56 | and are most of all black students were actually |
---|
0:19:00 | the titanium believable that the machine could have anything like but |
---|
0:19:05 | no effect on very clear people responded that wants to that's the |
---|
0:19:10 | the actually believe that the machine thing |
---|
0:19:21 | so we don't have instances of battery hundreds of people |
---|
0:19:26 | what we did not particularly if they were different from the a believable back to |
---|
0:19:32 | you |
---|
0:19:33 | really of the reciprocity affect and reciprocity effects or and we would like to these |
---|
0:19:40 | will store |
---|
0:19:45 | first |
---|
0:20:25 | so great question initially our work but only the shaded up to the |
---|
0:20:32 | and we didn't really ask questions but then people never school us because they just |
---|
0:20:37 | i know what to talk about rip of order |
---|
0:20:40 | the initiative to do this architecture may be kept asking questions and we might get |
---|
0:20:46 | one and i started do not include a direct responses to question |
---|
0:20:49 | so even something like are not clearly a i even things like i did see |
---|
0:20:55 | the movie we don't count those that self disclosure available they technically are giving information |
---|
0:20:59 | because they're |
---|
0:21:00 | just on sorting the question but the bare minimum of what require |
---|
0:21:04 | so we only consider things with a probation it wasn't necessary as a as close |
---|
0:21:09 | to |
---|
0:21:10 | and that of the data collection |
---|
0:21:14 | last |
---|
0:21:32 | do you mean that is us |
---|
0:22:05 | i think there had so the people are currently work |
---|
0:22:10 | okay |
---|
0:22:12 | okay i think |
---|