transcript_MAN_annot09.txt

(PERSON7) Aha, so now I can do that.
(PERSON6) Yeah, right.
(PERSON7) Now I'm recording already.
Fin, aha I can see that.
And maybe I'll ask you before anybody else is ehm connecting,
ehm how is it – how does it work with the link?
Do you create a next different link every time with – have a meeting?
(PERSON6) No, no, this is the same link you can find in the –
if you have accepted ehm that ehm meeting invite, it is the same link.
(PERSON7) mmhm, so that's –
(PERSON6) Meetings only.
(PERSON7) Yes.
So this [ORGANIZATION2] link will also – ehm will also function the next time we ehm have the coffee meeting.
(PERSON6) Yea, yea, yea.
It is the – it is the rolling [ORGANIZATION2] link,
we can use it ehm like for – for all of our meetings, so –
(PERSON7) Yes, okay, nice.
(PERSON6) For all our coffee meetings.
(PERSON7) Perfect, I just –
(PERSON6) So I'll also put – I'll also put the [ORGANIZATION2] link as well as the [ORGANIZATION2] link, so <unintelligible/> is suitable for us.
We can <unintelligible/>
So in the same – same meeting invite you can find <unintelligible/> links.
(PERSON7) Okay, mmhm.
Thank you, thank you.
(PERSON6) Okay.
(PERSON7) And now let's wait before [PERSON4] or – a- and [PERSON2] are connecting.
(PERSON6) mmhm
<unintelligible/> said that he had another meeting ehm before ehm a- 
so from Doodle so – so – th- th- an hour before – before that.
Maybe he didn't finish yet, I'm not sure. 
(PERSON6) Okay.
So, [PERSON7], so <unintelligible/> let discuss some things, okay?
(PERSON7) mmhm mmhm
(PERSON6) So, I – I wanted to note that –
ehm so, I didn't know <unintelligible/> –
it was like –
I don't think there is <unintelligible/> there is the – ehm everything I <unintelligible/> clean manually corrected transcript as well as <unintelligible/>.
Is that correct?
<unintelligible/>
(PERSON7) So our output is ma- manually corrected minutes and manually corrected transcript.
Is that what you asked?
(PERSON6) I – I – I want to know that how do you have <unintelligible/>?
Like, how do the <unintelligible/>
(PERSON7) ehm Okay.
ehm Did you see the document Minuting annotation guidelines?
Probably –
(PERSON6) Yes.
(PERSON7) If I ehm –
if we look at it together and I will just ehm detailise some things that you want to know ehm –
so ehm, ehm I will send you a link to minutes annotation, so that you can open it now.
(PERSON6) Okay.
(PERSON7) ehm Do we have some chat here?
Yes.
(PERSON6) ehm Maybe you can send it to the hangnote.
(PERSON7) ehm I have already sent it by chat, but I also can sent it by hangnotes.
(PERSON6) Okay.
(PERSON7) ehm <unintelligible/> is right that [ORGANIZATION2] is not working for <unintelligible/>.
How is it possible?
(PERSON6) Oh, yeah, how is it possible?
(PERSON7) So how is it possible that some link is working and another doesn't?
(PERSON6) I don't know.
But [ORGANIZATION2] is like, you know, much bigger than [ORGANIZATION2] meet.
(PERSON7) I also think –
(PERSON6) And we have problems.
(PERSON7) And especially I can hear much better with [ORGANIZATION2] than with [ORGANIZATION2] meet.
(PERSON6) Right, right, exactly.
(PERSON7) One more thing is – is that they want to make it ehm –
so that – so we sh- we have to pay for it beginning till next month or so.
(PERSON6) Oh, <unintelligible/>?
(PERSON7) [ORGANIZATION2] meet.
It's not going to be free any more.
(PERSON6) Oh, it is already free for one hour.
(PERSON7) Yes.
(PERSON6) But more then one hour you have to pay.
(PERSON7) Oh, one hour, okay.
It's not <unintelligible/> actually, okay.
ehm So –
(PERSON6) But [ORGANIZATION2] meet is also free for fourty minutes, but I want to <unintelligible/> [PERSON7], how like ehm [PERSON6] ehm like be have got like the paid for <unintelligible/> [ORGANIZATION2].
Because <unintelligible/> [PERSON6] seems that [ORGANIZATION2] meeting link, it is obvious the – like <unintelligible/>
(PERSON7) ehm We have the paid version of [ORGANIZATION2] for our faculty.
ehm But –
(PERSON6) Oh –
(PERSON7) There are some [ORGANIZATION2]s that we prepaid, but ehm I'm not sure if it's a good option for lo- for our coffee m- coffee meetings.
Because ehm that's kind of for everybody and ehm if we have not cross with other ehm – with other courses that are going on on ehm – at our faculty.
(PERSON6) Okay, okay.
<unintelligible/>
(PERSON7) You really have to follow if nobody else is – s- needs this hour for their teaching,
so it's better to have something outside.
(PERSON6) So [PERSON7], could you – could you respond to <unintelligible/>, like if he can reensolve [ORGANIZATION2] and then join, because in [ORGANIZATION2] meet it is really difficult to hear.
<typing/>
(PERSON7) mmhm
(PERSON6) Okay.
ehm Well, ehm so, what we – ehm we begin this ehm data collection.
ehm I will try ehm to make a very ehm short presentation how I do that.
ehm So, we have automatic speech reconstruction transcripts.
So we have transcripts with really bad automatic quality, which I send to the annotator and ask him to correct the data.
And I also send him ehm the audio file or video file ehm via FileSender.
ehm This is some service of our university that I can send ehm big ehm attachements.
I send a big attachement via some server and link to it,
and in the attachement of the e-mail i send him the pre-processed automatic speech reconstructed – speech reconstruction transcript.
ehm So this is the stage one, transcript annotation, and he listens to the audio and corrects the transcript.
After that ehm he ehm writes ehm from stretch the minutes and send them to me.
This is –
(PERSON6) <unintelligible/> Hold on.
So when you send that –
so <unintelligible/> so what the annotator does is that he or she listens to the audio, and also goes to the automatic speech – ehm speech transcript.
(PERSON7) Yes.
(PERSON6) ehm And then?
He or she corrects that?
(PERSON7) Yes.
(PERSON6) Also based on how they understand it?
<unintelligible/>, am I right?
(PERSON7) He creates the minutes, yes.
ehm Mostly I –
(PERSON6) Okay.
(PERSON7) <unintelligible/> not to give them the original minutes.
So if they really ask for that, I can – I sometimes send them ehm just for –
once <unintelligible/> they look like, but mostly –
(PERSON6) But <unintelligible/>
What is the original minutes?
(PERSON7) This is ehm just what [PERSON6] created duri-
the agenda for the minutes.
(PERSON6) So this is the – like – <unintelligible/> by the participant, great.
(PERSON7) Yes.
This is mostly – it – this is what was created by [PERSON6] ehm within our [PROJECT1] minutes – [PROJECT1] meetings.
(PERSON7) So there are ehm –
if you look at the data, there are a lot of ehm docs there and data copy pasted from ehm [ORGANIZATION2] documents.
ehm And the documents ehm contain original agenda of the meeting, which was created by [PERSON6] or [PERSON3] before the meeting and partially ehm s- <unintelligible/> extended during the meetings, or sometimes after the meeting.
But this is kind of work version of what really was.
(PERSON6) Okay, so I would like to ask you now –
So, by this original meetings, what <unintelligible/> [PERSON6] and [PERSON3] has created, so it is just for reference of the annotators.
So <unintelligible/> –
(PERSON7) This is just for what?
Sorry, once more please.
(PERSON6) This is just for the reference of the annotators.
ehm Do you think it has any practical significance to our – ehm our <unintelligible/> or media task?
(PERSON7) I think it has, it may have.
It's not –
so this is also the output.
<unintelligible/> doesn't have the full representation as needed yet, but it can be changed to this presentation and it also reflects the essence of the meeting.
so I cannot actually say that they are worse that the min- minutes created by annotators.
It is actually a questions, because there are some drawbacks of those originnal minutes, because ehm they do not contain some information or they may contain something that was not discussed during the meeting.
But on the other hand they are rea ehm, so they contain real names ehm and they don't have errors in – ehm in the content.
So actually they're not worse.
We can put them together with other ehm – with transcripts and get more, just get more –
(PERSON6) So, not all the meetings has the original minutes, right?
(PERSON7) ehm Yes, ehm we try to get original minutes from the minu- meeting organisers, but sometimes ehm it may be very short for example, for –
Some group of meetings, for example for [PROJECT2], ehm the minutes are very very short and they contain just one line [PROJECT2] book.
And they discuss the book, not detailise it anymore.
(PERSON6) Okay.
Okay, so, so ask <unintelligible/> line is good.
So, you send this via file sender, and they <unintelligible/> you back the manually corrected ASR?
And the minute for <unintelligible/>, right?
(PERSON7) Yes.
(PERSON6) Okay. 
(PERSON7) And I put in –
I put it to the server.
(PERSON6) You put it <unintelligible/>.
So now, how did it work?
So did the teams <unintelligible/> directions?
Now how did it work.
(PERSON7) ehm Now it works the same, we din't – we didn't change it yet.
But after we change it –
So we didn't change anything yet, so actually ehm, probably you have to change it.
(PERSON6) ehm No.
I want to say that support after the team watch them, where from today look the <unintelligible/>.
Today we look into the server, or should I have to be in touch with the annotator?
(PERSON7) This is the question we discussed today in the morning.
ehm So [PERSON6] proposes that annotators do that themselves.
We proposed that they will send the data to us.
ehm So I still kind of –
so at the moment I do not have any idea hw to technically work with the variant that annotators send the data via GateHub.
[PERSON6] – [PERSON6] is very much for that so that we really do that GateHub team and they can push and pull the data to the GateHub.
ehm But at the moment I'm not ready to say how it should be organised, because I have to give them ehm tasks ehm s-
I have to say them'Now you ehm write this and that – this meeting' ehm, I have to correct it, so I have to look if their transcript and the minutes are ehm okay.
I have to provide – provide them with the audio and video file,
and this is an extra action that should be done apart from GateHub, because it is a hard file.
(PERSON6) Okay.
(PERSON2) ehm Sorry to interupt you [PERSON7], I think that [PERSON6] already decided not to include – not to ehm let them <parallel_talk/> Yes. <parallel_talk/> work with GeteHub, because it deleted the – the team.
It deleted the team <unintelligible/>.
(PERSON7) Yes, but after <unintelligible/> –
(PERSON2) But I think we – I think we agreed that they will keep communicating with you in the same way and that <unintelligible/> will manage log it up –
(PERSON7) Well, well –
(PERSON2) <unintelligible/> what you will –
(PERSON7) [PERSON2], not really, because afterwards, after that meeting, he said me many times h- that he still s- sees that it should be maybe changed.
So maybe not today but he still sees the option ehm better that this teams would be created again, <unintelligible/>.
(PERSON2) Okay, <laugh/> so then we need another meeting I guess.
then we need another meeting with [PERSON6].
(PERSON7) Not too long, but he- s kind of ehm –
we ehm spoke about that GateHub for a while ehm after tha- afterwards.
(PERSON2) Okay.
(PERSON7) He – wha- he didn't try to persuade me to made these teams, but in his vision this teams still figurated somehow.
(PERSON2) Okay, okay.
(PERSON7) Probably we should –
(PERSON2) So, [PERSON7] –
(PERSON7) Not exclude this option for the moment, ehm but –
(PERSON2) Okay, so, [PERSON7], I – yeah I – so – so this is like you <unintelligible/> to upload the minute <unintelligible/> correct, so this is actually a list work for us.
So we sould be absolutely happy with that.
But the thing is, or the only thing we worry is, because since you said <unintelligible/> technical people, so if something goes wrong <unintelligible/>, otherwise we're absolutely fine with that.
(PERSON7) What is absolutely fine?
(PERSON2) Like if the annotators up- upload the minute themselves, so –
(PERSON7) Actually I'm not really fine with that at the moment, because for me – ehm I should think it over and I didn't do it yet.
So I don't know how would I have the control o- over annotators.
The communication.
Because I have to send them what all the files manually, otherwise –
so I don't know how to do that.
I really have to upload the data and it takes time.
(PERSON2) I – I think the same [PERSON7], I absolutely agree with you.
I would not ask this persons to push in GateHub <parallel_talk/> Yes. <parallel_talk/> in the final repository ehm to push there anything.
So I would also gain ehm – say that ehm I prefer this – the same strategy that you're using, <unintelligible/> sending the data –
(PERSON7) So probably let's – yes.
Let's not do this team for the moment.
It could be changed, I think it could –
so, wh- I and [PERSON6] still thought about this option is not – not because of the present situation.
Now it's absolutely okay, that we just send them the data.
But he thought, overal the situation where they make alignments and it takes more time and ehm somehow it can – it – it can –
it is connected with alignments.
ehm
(PERSON2) Okay, okay.
(PERSON7) So let's not think about it at the moment if possible.
And if needed we would ehm kind of build it on the- ehm – on the existing system.
So at the moment I would prefer to have the contact with annotators, but ehm actually I'm really <parallel_talk/> Wonderful. <parallel_talk/> good in that, so rather use <unintelligible/>.
(PERSON2) Okay. 
(PERSON6) So, <unintelligible/> [PERSON7], so ehm – so ehm – let – [PERSON2] said in the morning –
I also like –
to some extent I thought and I – I think I also believe to some extent –
so – ehm <unintelligible/> we have an actually manual evaluation <unintelligible/>.
What is the – like, what is the vision <unintelligible/>.
(PERSON7) Aou want me to repeat why you need the alignment?
Why we need the al-
(PERSON6) No, no.
I – I need the al-
I – understood we need the alignment, but ehm I was talk <unintelligible/>, so we must have a measure or a <unintelligible/> by which we can use the alignment to produce something.
(PERSON7) mmhm
(PERSON6) Great, so, we do alignment, <unintelligible/> fine, the manual <unintelligible/> are done, but what is the final output?
It should be some <unintelligible/> files and some nubers.
(PERSON7) Final oi- output of what.
(PERSON6) Final output of the alignment.
(PERSON7) ehm <unintelligible/>, could I ask you, please, now, if you could write this very question very explicitly somewhere into – into the agenda.
(PERSON6) Okay.
(PERSON7) Because I think ehm there should be a total understand– understanding between us what do you mean.
I'm not sure ehm –
first I'm not sure that I understand what you ask and second I'm not sure  – I'm not sure I can answer.
I'd rather think that answer should go from other source.
(PERSON6) Okay.
Okay, so I'm <unintelligible/> typing in that – ehm in that –
(PERSON7) Just form the question.
(PERSON6) – [ORGANIZATION2] doc.
(PERSON7) mmhm
(PERSON6) So how are we going to use the alignment in evaluation?
<typing/>
What is the measure or mid-check of this activity?
(PERSON7) mmhm Well, I – ehm now I understand it.
I think it's a very good question and ehm I don't think that we have answered it yet.
So we as-
[PERSON6] has kind of vague idea that it helps and so have I.
ehm So this w- ehm
so we have some ehm experimen- experience with other tasks where such alignment helped somehow.
And we have the intuition ehm based on some experiment- experiments with other data, that it would help.
Probably I'm a little bit sceptic now, but ehm I – I really think that at the moment it is not solved.
So this is what – ehm
if you would ask [PERSON6] this question he would probably say that ehm <unintelligible/> think about it and ehm make o ehm or invent the measure or metric for – for that.
So I've like – I – I would pick up from this here, so the thing is that ehm meeting or coming up with that measure want to be that much busy.
Like I'm just talking form <unintelligible/> point of view.
So, probably we have to submit the proposal ehm within the first week of November.
Okay, so, it –
I do see th- this coming within October, that I'll be <unintelligible/> measure and that is accepted that, okay?
This is a manual measure for ehm automatic minority.
So the thing is that this is <unintelligible/>, right?
So, do you think that it could be <unintelligible/> thing?
Because ehm if we ehm like –
I don't want a <unintelligible/> that –
suppose we say about –
(PERSON7) Sorry, sorry, I lost the u- lost the idea.
ehm Do you think that we should include what, evaluation measure?
(PERSON6) Do you think this alignment –
(PERSON7) Or alignment?
(PERSON6) First, alignment first, manual evaluation measure –
(PERSON7) Okay –
(PERSON6) Should be a part of the shared task?
(PERSON7) ehm <unintelligible/> I don't think about it.
I don't know, I have no idea.
So I really don't know.
I think this is the <unintelligible/> you should discuss in more detail with [PERSON6].
(PERSON6) mmhm
(PERSON7) Because –
(PERSON6) So I <unintelligible/> this points.
(PERSON7) mmhm 
(PERSON6) So this –
I would say –
(PERSON7) I have never had any experiments with shared task and measures and so on.
So I – what I am saying is kind of my intuition.
ehm I <parallel_talk/> Okay. <parallel_talk/> cannot answer.
(PERSON6) mmhm Okay.
(PERSON7) Rather ehm –
go-
either think with [PERSON4] about ehm – about that or ehm think ehm with [PERSON4], [PERSON2] and [PERSON6], not me, I don't know.
(PERSON6) mmhm Yeah, okay, okay.
Okay, thank you.
ehm Okay, so- suppose they does speak a scenario that we don't come up with a manual evaluation measure, before the shared task, so, what options we have.
So the options that we have is – we have the transcript, which is manually corrected, we have the minute, which is the <unintelligible/>,  contains the <unintelligible/> summary.
And then ehm we reuse the data and we evaluate the submissions based on automatic measure.
(PERSON7) ehm Do we –
where do we get the automatic measure?
(PERSON6) There are many automatic measures for summarisation.
(PERSON7) mmhm But <unintelligible/> none of them works as [PERSON4] realised.
(PERSON6) ehm How many ehm do you have tried?
(PERSON7) How many what?
(PERSON6) How many – how many automatic measures have you ehm tried?
(PERSON7) ehm Ask [PERSON4].
(PERSON6) Okay.
(PERSON7) Let –
ehm so in summarising paper on that <unintelligible/> maybe [PERSON2] also knows, none of them really works well.
(PERSON2) Well I wouldn't say none of them really works well, I would say that they are kind of deceptive in some cases like they tch- they are very high – very high <unintelligible/> –
(PERSON7) Yes mmhm.
(PERSON2) And then somebody <unintelligible/> just <unintelligible/> repeated this that – this that – this that, so you <unintelligible/>.
So they must be c- compo- ehm – like associated with the human score.
Like decives them, we first add them, because we – they are very easy to compute.
But we add another human evaluation score, that was the point of ehm th- that I wrote the protocol, like try to come up with some tips.
So, we need the na- next try human evaluation, decides them, we compute them, but we add the number of human evaluation, so that we can ehm be sure that we have good  quality.
That's the point.
(PERSON6) I agree with you.
But that human evaluation <unintelligible/> should be like very subjective and <unintelligible/>, right?
(PERSON2) mmhm No, it's not very subjective, it's like –
we I – I think –
did you read the document that I wrote?
The protocol I wrote?
(PERSON6) Yes, yes.
(PERSON2) There is how most of the like evaluations are done in the community, in the literature.
Like have a summary or a text and we compare it with the reference and we have the annotators th- to compare it.
And they like try to assess different aspects like <unintelligible/> reada- readibility, the quality etc., and they put scores of that.
And then we have this five stars for rating – <unintelligible/> rating scale, from one to five stars.
And then we abbregate, we wait this ehm – let say this aspects, and we come up with the – with the final score.
This is the most common and easy way to – to do it you know.
That's why I proposed that.
But then [PERSON6] has his own idea about this alignments and I don't really know, I know nothing about, so that's it.
(PERSON6) Yeah, yeah.
I – I – I got – I go- got the multimission of ehm the <unintelligible/> meeting script, but still that is like the <unintelligible/> that [PERSON6] ehm is targetting to achieve ehm would you be more objective?
And the liker giving skills, okay, but still I would believe that it is subjective.
(PERSON2) ehm Yes, I mean –
(PERSON6) Right.
<unintelligible/>
(PERSON2) It's human based and every – every human evaluation is subjective, so –
<unintelligible/>
Everyone is doing it that way, so –
I mean most of the research evaluat- human evaluation in nature language generation is done that way, so-
(PERSON6) Okay, wonderful.
<typing/>
<unintelligible/>
Okay, so, like –
Okay, so the thing is that if we don't come up ehm with the alignment based manual evaluation <unintelligible/>, we will do as the normally the other energy task <unintelligible/> ehm entries, right?
(PERSON2) I'm really sorry, I didn't follow you.
(PERSON6) Yeah, so <parallel_talk/> So – <parallel_talk/> if we – if we are like – if we are not able to come up with the alignment ehm, if we are not able to come up with the alignment based ehm manual evaluation measure, –
(PERSON2) Yes.
(PERSON6) We go with – we go with the <unintelligible/>, the <unintelligible/>
(PERSON2) Yes, sure, sure.
Yes, why not, yes.
That's – yes, okay, why not.
(PERSON7) Okay, so may I – may I summarise it, please?
So ehm –
(PERSON2) Yeah.
(PERSON7) So ehm really ehm we want to use ehm alignment annotation for summarisation, this is variant A.
But we have ehm little time and very probably we don't have to ehm prepare this alignments.
And in this case we follow variant B, and this is using ehm human evaluation like readibility and so on, not using the alignment.
(PERSON6) Right.
(PERSON7) So –
(PERSON6) mmhm Yes.
Yes, so ehm – so the thing is that ehm my ehm <unintelligible/> before November it would be difficult to come up with the alignment this score, so ehm for the past shared task that we have way to propose <unintelligible/> speech,
I think we should go whatever is available and whatever is better to us.
And then maybe we would also like do some research on this and we come with some beautiful matrix,
and then maybe the second ehm – you know the second event <unintelligible/> we can come up with this ehm new <unintelligible/> – new techniques that we want to do.
Halo?
(PERSON2) So this – this –
you were talking to [PERSON7] or to me?
(PERSON6) ehm I'm talking to you <unintelligible/>.
(PERSON7) <laugh/>
(PERSON2) So, <laugh/> I'm really confused.
(PERSON7) <laugh/>
Probably yes.
He just – you are just trying to summarise what we did and what we said, yes, yes.
(PERSON6) I'm just worried about like what ehm the proposal should look like, because ehm like for variant B there are still like some cloud on that.
We are not very clear.
So, ehm I just want that the proposal should be like very strong, and whatever we say shoould have a ground – grounded this is to that.
<unintelligible/> so, I now loose –
I know there are many autoatic measures for summarisation so if we like just put those ehm in the proposal ehm no one we question?
And we – if we like say that we are going to human evaluation using the ehm like pst- like the reading skill or how they <unintelligible/>, no one we yuestion?
But if we are not believed here about the alignment <unintelligible/>, so that's it.
That's why my <unintelligible/> nice.
(PERSON7) ehm I kind of don't know what to say.
So ehm, I don't knw what you propose.
You propose ehm directly to go plan – plan B, yes?
(PERSON6) No, B – we – like we have plan B as our backup.
If varian A doesn't become clear to us before the proposal submission deadline, we move in variant B.
If variant A is very clear, its concensus is accepted by all of us, we obviously go with variant B.
(PERSON7) mmhm Okay. 
(PERSON6) Yeah.
Okay, okay.
Okay, so, yeah so I think – I think <unintelligible/> no – <unintelligible/> –
[PERSON7] you wanted to discuss something?
I have understood what is <unintelligible/> that [PERSON6] was speaking about.
Yeah, it's how we should do the shared task and –
(PERSON7) ehm <parallel_talk/> Yeah, so – <parallel_talk/> so, ehm what I wanted to discuss ehm is probably ehm the format of the basic sample trial.
But this is – I think we should read with this – that format before [PERSON1] ehm appears and discuss it with [PERSON1].
(PERSON6) mmhm
(PERSON7) So this –
my – my point of how to create basic samples should be discussed with [PERSON1] before we ehm go on.
So for now I don't prepare any basic samples more.
And then I ehm –
wh- when [PERSON1] appears ehm I will discussed – ehm discuss with him, if his annotation tool can work with the <unintelligible/> samples,
and then prepare ehm the second ehm version of basic samples.
Point?
And ehm, as for the data format, this is also a point with [PERSON2] ehm – 
(PERSON2) As for the data format I'm already clear, like it's what we discussed like the three – three sections in each file.
The first section is meta-data, you put on meta-data there,
and you're right in ehm <unintelligible/> brackets meta-data.
And then, the second distrust it, as I each <unintelligible/>, so the second and the third are crossed in the summary.
And in the summary you can keep the bullets, but only with dashes.
(PERSON7) Yes.
(PERSON2) <unintelligible/> one dash, two dashes etc.
(PERSON7) Yes.
(PERSON2) So this is – I think this is the ehm –
so it's almost like I said, but you can –
I said to remove the meta-data,
you can put them with ehm – with a label in your <unintelligible/> brackets –
(PERSON7) Yes, meta-data and dashes, dashes is important.
(PERSON2) So it's – it's almost the same, so the format is stable <unintelligible/> –
(PERSON7) Yes, this is –
Well, I think I will have to discuss it with you before I begin once more about some details.
(PERSON2) Yes, sure, because – 
yes, <unintelligible/>
(PERSON7) But, yes yes, but –
(PERSON2) But I think this is a –
because this way you put all the information, you put practically put on all information, because you have the meta-data <unintelligible/> the summary.
There is nothing <unintelligible/> –
<unintelligible/>
(PERSON7) Yes, yes.
So it's good, isn't it?
It was not my point.
(PERSON2) You deserve the hierarchy of the bullets ehm, but only with <unintelligible/> recognisable, because <unintelligible/> is not recognisable, so.
(PERSON7) Yes, yes, mmhm.
(PERSON2) I think for this we are okay, but we didn't have time in the morning to discuss about some other details, like ehm anonymisation of – anonymi- anonymisation of the names of this annotators.
We need – we need to encall them, so you have to prepare a list with their real full names and their annot01, annot02, annot03 and that you have to put in the ehm folders.
So, you will have to rename [PERSON5].
You will keep a list, where [PERSON5] is like [PERSON5] what – I don't know is her name and then you put annot01, as a sample example.
And then you will use the zer- annot01 as the name to the folder structure that you – that we have, so you will have to rename the folder structure.
(PERSON7) [PERSON2], lwt's do it –
so, the question <unintelligible/> do it when we do it who does it.
ehm So, an I –
(PERSON2) What do you mean?
What – ehm what – what, like renaming in the structure in the directory or the list, or what do you –
(PERSON7) So, now we are ehm looki-
we are ehm kind of ehm changing ehm the data storage from Sorbow to GateHub repository.
(PERSON2) ehm No, we will keep both of them.
We'll keep the <unintelligible/> –
(PERSON7) Yes, I know, but st-
yes.
But we are kind of centralizing it and we're making server as one of mocking versions of GateHub repository, so who is doig all of that?
(PERSON2) Once again, please, I couldn't hear you.
Where are restoring <unintelligible/> –
(PERSON7) Who – who does all that GateHub ehm things?
(PERSON2) For the GateHub it's only pushes, I <unintelligible/> do it.
there is nothing to do in the GateHub for you.
You will just walk in the cluster.
What I was saying is that you or I or together we will have to rename the names in the cluster, all the names.
(PERSON7) Yes, yes –
(PERSON2) From [PERSON5] – from [PERSON5] you will put annot01, like anonymise the name.
(PERSON7) ehm When?
Before ehm a create in GateHub or it doesn't – ehm or it doesn't –
(PERSON2) Before pushing, before pushing to Gatehub.
(PERSON7) Before pushing?
This was my question.
(PERSON2) Yes, yes.
(PERSON7) So –
(PERSON2) Yes, sure, before pushing.
Because in GateHub it has to appear as anonymous, as annot- annot01.
(PERSON7) <another_language>aha</another_language> Yeah, so before pushing.
I have to change all the names,
so it means, that for each annotator I have to change the name form [PERSON5]1, from [ANNOTATOR2] to annot2 –
(PERSON2) Yes.
(PERSON7) Somebody –
so – and if I –
(PERSON2) Yes, so you create the file – you create the file the document that – internal – that you have access to, or even a notebook let say, a paper notebook,
and you have to keep the – the names of this collaborators or annotators, their full names –
(PERSON7) Yes, this is clear.
(PERSON2) And an anonymous name that is annot01, annot02, etc., yeah?
and you will <unintelligible/> record of this list –
(PERSON7) Perfect, so this is okay, yes.
(PERSON2) <unintelligible/> record of this list when you name the folders with their collaborators.
(PERSON7) Yes, so this is – this is what –
(PERSON2) <unintelligible/>
(PERSON7) This is what I like.
If I – if I am allowed to do it so, I like it so.
Because ehm I'm not able at the moment, I'm not sure if I can be – if I can make <unintelligible/> but I – at the moment it is very complicated for me to anonymise ehm all of the files to – so to name general names.
I have to keep the names – this final names of the files recognisable.
So it may – it should not be the name of the annotator, but it should be some ID that I can ehm dete- that I can determine where to put it. 
(PERSON2) Yes, I told you, annot01 is like –
(PERSON7) Yes, yes, in this case it is a <unintelligible/>, so nw my task is to rename the files.
(PERSON2) yes, but before that I also have –
I mean, I also propose some other things but you didn't have time to discuss them, so, I'm not sure –
(PERSON7) So, let's meet today, [PERSON2], let's meet today in person?
<unintelligible/> to <unintelligible/> more –
(PERSON2) [PERSON6] would be – better to have also [PERSON6],
I don't know, when he's available now, or tomorrow, or –
so maybe it's better to set up another meeting, like today –
like one of the day in the morning?
So, I think [PERSON6] –
(PERSON7) So, just <unintelligible/> –
okay, okay.
So let's write – write [PERSON6] when he has time, so –
(PERSON2) Yes it would be very good.
(PERSON7) Or I can <unintelligible/> Friday and <unintelligible/> again,
I have web seminars <unintelligible/> Thursday –
(PERSON2) Probably, tomorrow morning or evenyou want today afternoon or some time?
It would be very good to have also [PERSON6] to seome things.
(PERSON7) Yes.
(PERSON2) Because I'm really worried about this team that he's proposing, so,
I <unintelligible/> disagree with [PERSON6] about several things, so we nedd to discuss <unintelligible/>.
(PERSON7) Yes, <unintelligible/> be everybody <unintelligible/> in person, yes, perf- 
ehm I agree.
(PERSON7) So ehm –
(PERSON2) So –
(PERSON7) I'll write him and you now and –
(PERSON2) Yes, try to – we try to catch him.
(PERSON7) <unintelligible/> meeting tomorrow morning.
(PERSON2) 
we try to catch him as soon as possible.
(PERSON7) Yes, not today, I am already at home – <parallel_talk/> <unintelligible/> <parallel_talk/>
What?
(PERSON2) Okay.
<unintelligible/> the meeting.
I will –
(PERSON2) Yes, sure.
(PERSON7) Okay. 
(PERSON6) Yes, <unintelligible/> something like the meeting of today, you were also present <unintelligible/> –
(PERSON2) Yeah.
(PERSON6) Repeat the one today.
(PERSON7) Okay.
(PERSON6) Okay, okay.
Okay, so now we're just –
like everything is ehm like almost sorted.
the thing is that we have to like wait for [PERSON1] <unintelligible/> to actually ehm <unintelligible/> that – <unintelligible/> this annotation tool composed for new <unintelligible/> idea.
(PERSON7) mmhm Yeah, [PERSON4] is writing.
I was waiting and tried install [ORGANIZATION2] again and with different e-mail adress.
Now it asking me meeting password.
Could someone please provide <unintelligible/> to my lock?
Do you let me know if something concers me?
He needs some password to <unintelligible/>, what does it mean?
Which password?
Meeting password, what is that?
(PERSON6) I think the meeting pass-
yeah, the meeting password is minuting, which is already there in the meeting <unintelligible/>
(PERSON7) Could you – could it – could you please send it to [PERSON4]?
(PERSON6) Yeah, yeah, <unintelligible/>.
(PERSON7) But anyway, I would rather probably finish it for today.
(PERSON6) mmhm Yeah, okay.
I too think so.
(PERSON7) mmhm
(PERSON6) [PERSON2], anything else you want to discuss?
(PERSON7) No.
(PERSON2) ehm No, I mean I definitely – with definitely <unintelligible/> with [PERSON6] to –
because today we didn't finish, like we left some issues without dicsussing them, so ehm –
We cannot really discuss anything without [PERSON6], because ehm –
(PERSON6) Exactly, yeah.
There are like decission points, decisions points has to take, so yeah.
(PERSON2) So, there is nothing to discuss today anymore.
Okay –
(PERSON7) So, have a nice evening, everybody.
(PERSON2) Okay, goodbye.
(PERSON6) Yeah.
(PERSON7) Bye bye.
(PERSON6) Goodbye, and stay <unintelligible/>.
Bye bye.
(PERSON7) You too. 
 
I'm not, you know, wanting is like, yes?
Yes, in this case, it is he.
So now my task is to return...