words
stringlengths 0
4.03k
| end_time
float64 0.46
2.95k
| speaker
stringclasses 4
values | start_time
float64 0
2.93k
| session_id
stringclasses 100
values |
|---|---|---|---|---|
the i and tf id have inverse document | 1,499.52 | speaker1 | 1,497.12 | session_0ed514d2 |
term frequency inverse document frequency i did do it once i do have a java class that does it for something i don't know whether it will work for this but all right | 1,508.95 | speaker2 | 1,500 | session_0ed514d2 |
because that will be really handy then we can actually test it on the user as opposed to just doing it and not using it because if you just have the segmentation that's great but we can't we can't compare it to the hand annotated the hand | 1,525.11 | speaker4 | 1,508.96 | session_0ed514d2 |
segmented tool yeah also keywords gives you an whole new type of search you do keyword search but you could do keyword search could be topic search and it could be the same thing instead of it would just search for keywords when they you tell them their to projection for topics but actually searching keywords | 1,540.5 | speaker2 | 1,525.12 | session_0ed514d2 |
for | 1,542.52 | speaker4 | 1,541.92 | session_0ed514d2 |
you i mean yeah | 1,548.59 | speaker2 | 1,542.6 | session_0ed514d2 |
that | 1,548.75 | speaker1 | 1,548.6 | session_0ed514d2 |
will work that' would be really nice | 1,549.96 | speaker4 | 1,548.76 | session_0ed514d2 |
i suppose even calculating the the whats his faces themselves wouldt be too much too long the easy bit is that's probably easier to calculate them based upon in their whole occurrences in the corpus than it is to calculate them per topic because you don't have to integrate as much yeah | 1,574.6 | speaker2 | 1,553.32 | session_0ed514d2 |
definitely | 1,574.9 | speaker4 | 1,574.64 | session_0ed514d2 |
information no you can you can do search without tfdf you just can't rank the search yeah | 1,585.31 | speaker2 | 1,575.76 | session_0ed514d2 |
that's why i'm thinking too we need to i mean we're just your stuff i no no no but that's | 1,590.91 | speaker1 | 1,585.32 | session_0ed514d2 |
isn't | 1,591.15 | speaker4 | 1,590.92 | session_0ed514d2 |
that what the idea was in the first place to rank rank the results so that sort of | 1,595.63 | speaker2 | 1,591.16 | session_0ed514d2 |
but then the main idea was more like to to speed it up speed the sarch up because nxd over over the whole of the corpus was just not feasible | 1,602.6 | speaker4 | 1,595.64 | session_0ed514d2 |
yeah but that w slow down ranking it won't slow down | 1,605.93 | speaker2 | 1,602.8 | session_0ed514d2 |
no but the general inverted file | 1,609.54 | speaker4 | 1,607.32 | session_0ed514d2 |
yeah it still uses an inverted file but it ranks the results by the amount nice extra yeah i thought that was part of it but yeah it doesn't matter if it's not | 1,618.22 | speaker2 | 1,610.64 | session_0ed514d2 |
yeah i wasn't sure | 1,619.45 | speaker4 | 1,618.4 | session_0ed514d2 |
don't think we had that in the the document no | 1,622.71 | speaker1 | 1,620.16 | session_0ed514d2 |
we didn so | 1,624.95 | speaker2 | 1,622.72 | session_0ed514d2 |
what it's only interim | 1,626.28 | speaker4 | 1,624.96 | session_0ed514d2 |
yeah exactly i know | 1,627.46 | speaker1 | 1,626.48 | session_0ed514d2 |
but yeah i guess if you if that's not part of it don't worry about it | 1,633.8 | speaker2 | 1,629.76 | session_0ed514d2 |
yeah i mean if you have time | 1,635.07 | speaker4 | 1,633.84 | session_0ed514d2 |
yeah i only | 1,636.11 | speaker1 | 1,635.08 | session_0ed514d2 |
going do this time anyway | 1,637.8 | speaker2 | 1,636.12 | session_0ed514d2 |
because i mean then then it would give you like if you search for a word it would give you the the meetings but also how often that that word occurred in that meeting that would be so useful | 1,647.35 | speaker4 | 1,638 | session_0ed514d2 |
it just give you a rank it would that was the whole point actually i this is your top one this is your bottom yeah yeah but | 1,654.12 | speaker2 | 1,647.36 | session_0ed514d2 |
the thing is we're just looking for when it happens in a meeting if they're looking for a particular term or something they just want to know if that term exists there and where does it exist and i want to see it yeah but | 1,662.83 | speaker1 | 1,654.48 | session_0ed514d2 |
it | 1,663.44 | speaker4 | 1,662.84 | session_0ed514d2 |
how informative that tfdf is an informative score is it | 1,667.44 | speaker2 | 1,664.12 | session_0ed514d2 |
if you have if you search for language and you return like you know basically all of the meetings you want to see where they actually spoke about it and where they just mentioned it once | 1,676.39 | speaker4 | 1,667.68 | session_0ed514d2 |
if | 1,676.47 | speaker2 | 1,676.4 | session_0ed514d2 |
it's like a two word term does a tfdf | 1,678.76 | speaker1 | 1,676.48 | session_0ed514d2 |
your compound down | 1,681.25 | speaker2 | 1,680.36 | session_0ed514d2 |
it's not a compound down it's just two words together like what in i don't know i don't know just something like language in edinburgh university yeah as a compound though | 1,693.86 | speaker1 | 1,681.84 | session_0ed514d2 |
like edingnynny day yeah and i can add it to me | 1,700.36 | speaker2 | 1,696.44 | session_0ed514d2 |
and | 1,703.07 | speaker1 | 1,702.96 | session_0ed514d2 |
it's my simple form it would do separate rank for each one each term you could make it more complicated than make it it's | 1,709.67 | speaker2 | 1,703.08 | session_0ed514d2 |
not make it more complicated | 1,710.92 | speaker1 | 1,709.68 | session_0ed514d2 |
and | 1,711.71 | speaker2 | 1,711.56 | session_0ed514d2 |
then add those up or something | 1,713.31 | speaker4 | 1,711.72 | session_0ed514d2 |
or what because combined thing is the way we were doing it we were just looking for the words period each word in the index well that's the way we were thinking about and then it's just saying were these two words in any of these documents and then if it was then we go closer and do the the nxt search | 1,727.7 | speaker1 | 1,713.44 | session_0ed514d2 |
yeah to | 1,729.75 | speaker2 | 1,728.56 | session_0ed514d2 |
look for the exact term or the regular expression or whatever guess you just do a sum | 1,733.31 | speaker1 | 1,729.76 | session_0ed514d2 |
of of the individual tfrdf for each term returned and that generally will be a bit crude but it will give you a score and the higher the more more informative each term is for each thing would give you it give you a high thing it's pretty crude anyway if it's just looking for if it's all it's going to do is look for separate oh because then it's going to go into the nxt sestion return or | 1,759.31 | speaker2 | 1,733.32 | session_0ed514d2 |
we choose the list of meetings that we wanted to search to do that yeah yeah that's true so yeah | 1,763.03 | speaker1 | 1,759.32 | session_0ed514d2 |
that's less crude | 1,764.12 | speaker2 | 1,763.04 | session_0ed514d2 |
like i'm just wondering if it's going to give us something cool then yeah absolutely but if it's sort of like if the nxt search has still got to be run to find these terms or these patterns yeah without doing any like | 1,777.46 | speaker1 | 1,766 | session_0ed514d2 |
wordb | 1,779.11 | speaker2 | 1,778.88 | session_0ed514d2 |
pairs or also like do wild cards work for something like that so if you're looking for a wireless wired wire blah blah blah you do wire with a star and that way actually that's a problem for me too so that's | 1,794.48 | speaker1 | 1,779.12 | session_0ed514d2 |
yeah yeah but the idea is that gives you an informity score how you combine that is | 1,800.54 | speaker2 | 1,795.76 | session_0ed514d2 |
is | 1,801.67 | speaker1 | 1,801.52 | session_0ed514d2 |
up to you i guess there's lots in the literature its work yeah okay if you there's a whole load about it in manning and shz they've got a whole chunk about so just ir isn't it's basic information okay they've got a big a good chapter on it if you haven't got it's on cognit yeah i've got | 1,820.06 | speaker2 | 1,801.68 | session_0ed514d2 |
somewhere p d f yeah | 1,824.84 | speaker1 | 1,821.2 | session_0ed514d2 |
me too hundreds of pdf was | 1,828.31 | speaker2 | 1,824.96 | session_0ed514d2 |
yeah do the pdf switch it to postscript switch it back to pdf | 1,830.98 | speaker1 | 1,828.32 | session_0ed514d2 |
okay sas | 1,837.76 | speaker2 | 1,836.48 | session_0ed514d2 |
o that might be good okay so to come back to the | 1,844.35 | speaker1 | 1,837.8 | session_0ed514d2 |
start screen | 1,845.1 | speaker4 | 1,844.36 | session_0ed514d2 |
okay | 1,846.28 | speaker1 | 1,845.68 | session_0ed514d2 |
sort | 1,847.07 | speaker2 | 1,846.92 | session_0ed514d2 |
of deing | 1,847.84 | speaker1 | 1,847.08 | session_0ed514d2 |
i'm i'm i'm i'm very unsure about what that should look like i mean we want to for the part where you can choose choose the the meeting what kind of information do you want to have about the meeting i mean the the longer name obviously if there is one because they don't all have a longer name what the users that spoke users that took part yeah then that's | 1,868.64 | speaker4 | 1,848.12 | session_0ed514d2 |
pretty | 1,869.83 | speaker1 | 1,869.56 | session_0ed514d2 |
much the same for the same group | 1,871.39 | speaker4 | 1,869.84 | session_0ed514d2 |
probably for the most part i mean there will be a set of people thating | 1,877.64 | speaker1 | 1,872.92 | session_0ed514d2 |
yeah just to give them an idea general idea of what there is and so they can pick a search | 1,884.51 | speaker4 | 1,878.76 | session_0ed514d2 |
for stuff or start working yeah i guess yeah i guess that's browse okay like search browse | 1,890.67 | speaker1 | 1,884.52 | session_0ed514d2 |
like | 1,893.54 | speaker2 | 1,893.4 | session_0ed514d2 |
if you're looking for something specific like is that what you're thinking like if they if they know exactly where they want to go they want to go to you know that meeting on december third and they just want to go there and see stuff then they can immediately go and brows yeah i want to look at the | 1,904.35 | speaker1 | 1,893.56 | session_0ed514d2 |
meeting where sons so spoke yeah of that know some sort of research group i want that first meeting look at that | 1,913.55 | speaker4 | 1,904.36 | session_0ed514d2 |
yeah yeah browse by meeting browse by bla browse by speaker like no i'm just thinking of different ways we can do that like different buttons on the top like if they're looking for particular person particular working group particular whatever i mean we just we could break things down like that to that level of detail if we want or we could | 1,937.44 | speaker1 | 1,913.92 | session_0ed514d2 |
yeah because we we just want to have a way of yeah because as soon as they've got loaded their first meeting they can browse through that quite nicely | 1,950.66 | speaker4 | 1,938.92 | session_0ed514d2 |
yeah but i'm just thinking like initially when they load up what are the range of things that they'd want to do yeah like if they're part of x working group they'll want to get in there do we want to save preferences what do you mean well if they're part of b d b working group they're gonna to want to look at the bdb ones and so default to that or i't know how are we going to know that | 1,975.06 | speaker1 | 1,952.32 | session_0ed514d2 |
based | 1,976.19 | speaker4 | 1,975.96 | session_0ed514d2 |
on what they did the last time | 1,977.23 | speaker1 | 1,976.2 | session_0ed514d2 |
have safe preferences yeah could have a safe preferences i guess | 1,983.12 | speaker2 | 1,979.12 | session_0ed514d2 |
but | 1,984.23 | speaker1 | 1,984.08 | session_0ed514d2 |
that would defy the whole point of browsing right | 1,986.06 | speaker4 | 1,984.24 | session_0ed514d2 |
if you only | 1,987.47 | speaker2 | 1,986.92 | session_0ed514d2 |
wanted to look at | 1,988.52 | speaker4 | 1,987.48 | session_0ed514d2 |
call it favorites then yeah you could have aroll down | 1,995.31 | speaker2 | 1,988.72 | session_0ed514d2 |
the list and find your favorite one i mean | 1,997.35 | speaker4 | 1,995.32 | session_0ed514d2 |
yeah it's not enough information not like | 1,999.19 | speaker2 | 1,997.36 | session_0ed514d2 |
they come back every day and do the same browsing | 2,001.8 | speaker4 | 1,999.2 | session_0ed514d2 |
i forget | 2,006.59 | speaker1 | 2,005.84 | session_0ed514d2 |
okay | 2,010.59 | speaker2 | 2,010.36 | session_0ed514d2 |
well | 2,010.75 | speaker4 | 2,010.6 | session_0ed514d2 |
no | 2,010.87 | speaker2 | 2,010.76 | session_0ed514d2 |
i'm | 2,011.03 | speaker4 | 2,010.88 | session_0ed514d2 |
just | 2,011.15 | speaker2 | 2,011.04 | session_0ed514d2 |
getting back to sort of what could be on the the startup screen like there's just a range of things we could do we could have like search options browse options i mean just i don't know just playing around with ideas and then we can tie this into our evaluation tasks and say well if you want to search for a meeting or search for a user in a particular meeting then do this blah blah blah so we can yeah oh right of you know guide them and's just like hey my task is to do this and there's a button perfect | 2,043.58 | speaker1 | 2,011.16 | session_0ed514d2 |
no | 2,045.08 | speaker4 | 2,044.48 | session_0ed514d2 |
i | 2,046.99 | speaker2 | 2,046.72 | session_0ed514d2 |
know | 2,047.6 | speaker4 | 2,047 | session_0ed514d2 |
yeah it would be quite a good if it has yeah | 2,049.55 | speaker2 | 2,048.04 | session_0ed514d2 |
i wasn't going to put any you know any any help in there any sort of explanation of how the browser works to be honest | 2,056.91 | speaker4 | 2,050.4 | session_0ed514d2 |
no | 2,057.27 | speaker2 | 2,056.92 | session_0ed514d2 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.