args-framework issueshttps://git.webis.de/code-research/arguana/args/args-framework/-/issues2023-03-02T16:48:24Zhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/113Query suggestion for demos2023-03-02T16:48:24ZJohannes Kieseljohannes.kiesel@uni-weimar.deQuery suggestion for demosSomething like in netspeak, but maybe less obvious.
![image](/uploads/a6d53a370e9a3a043d86f31077b80d74/image.png)Something like in netspeak, but maybe less obvious.
![image](/uploads/a6d53a370e9a3a043d86f31077b80d74/image.png)Github Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/111Integrate MARGOT2022-09-08T11:12:26ZJohannes Kieseljohannes.kiesel@uni-weimar.deIntegrate MARGOTWeb service for argument mining from Uni Bologna:
http://margot.disi.unibo.it/#Web service for argument mining from Uni Bologna:
http://margot.disi.unibo.it/#https://git.webis.de/code-research/arguana/args/args-framework/-/issues/108Add publications to about page2023-02-27T08:35:13ZJohannes Kieseljohannes.kiesel@uni-weimar.deAdd publications to about pagehttps://webis.de/publications.html?q=keywords%3Aargshttps://webis.de/publications.html?q=keywords%3AargsGithub Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/105JSONL for args me corpus2023-02-27T08:36:05ZJohannes Kieseljohannes.kiesel@uni-weimar.deJSONL for args me corpusWe should provide future corpus version as JSONL for easier access.
`cat idebate.json | jq -c '.arguments[]' > idebate.jsonl`We should provide future corpus version as JSONL for easier access.
`cat idebate.json | jq -c '.arguments[]' > idebate.jsonl`Github Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/104Parliamentary and Quotes: Read CSV in Mining2023-02-27T08:36:22ZJohannes Kieseljohannes.kiesel@uni-weimar.deParliamentary and Quotes: Read CSV in MiningAcquisition just downloads the datasetAcquisition just downloads the datasetGithub Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/103Use display width for shortening titles2023-02-27T08:46:51ZJohannes Kieseljohannes.kiesel@uni-weimar.deUse display width for shortening titles"Append ellipsis when text overflows its containing element"
https://caniuse.com/?search=ellipsis
```
text-overflow: ellipsis;
white-space: nowrap;
overflow: hidden;
display: inline-block;
max-width: 100%;
```"Append ellipsis when text overflows its containing element"
https://caniuse.com/?search=ellipsis
```
text-overflow: ellipsis;
white-space: nowrap;
overflow: hidden;
display: inline-block;
max-width: 100%;
```Github Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/101Add paragraph premise annotations2023-02-27T08:36:54ZJohannes Kieseljohannes.kiesel@uni-weimar.deAdd paragraph premise annotationsBased on generic HTML (p, br), but first have a look at the debate portals.Based on generic HTML (p, br), but first have a look at the debate portals.Github Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/100Add git hash to build in shadowjar2023-02-27T08:37:46ZJohannes Kieseljohannes.kiesel@uni-weimar.deAdd git hash to build in shadowjarThis could allow us to have the API documentation link to the respective version in our gitlab.This could allow us to have the API documentation link to the respective version in our gitlab.Github Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/97Investigate and prevent ID duplicates2023-02-27T08:39:04ZJohannes Kieseljohannes.kiesel@uni-weimar.deInvestigate and prevent ID duplicatesIt seems there are arguments with the same ID in the dataset.
Some examples:
- Sed086351-A19b607d6
- Se511ec5-A4a6c46b4It seems there are arguments with the same ID in the dataset.
Some examples:
- Sed086351-A19b607d6
- Se511ec5-A4a6c46b4Github TransitionJohannes Kieseljohannes.kiesel@uni-weimar.deJohannes Kieseljohannes.kiesel@uni-weimar.dehttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/96Integrate conclugen as an alternative summarizer2021-05-21T09:37:55ZJohannes Kieseljohannes.kiesel@uni-weimar.deIntegrate conclugen as an alternative summarizer@fumi5146 has prepared the docker image. This should get a cache like the snippet generator: https://git.webis.de/code-generic/code-admin-knowledge-base/-/blob/master/services/args/snippets-args-me.k8s.yaml
```
docker run --name=conclug...@fumi5146 has prepared the docker image. This should get a cache like the snippet generator: https://git.webis.de/code-generic/code-admin-knowledge-base/-/blob/master/services/args/snippets-args-me.k8s.yaml
```
docker run --name=conclugen -p 8080:8080 webis/conclugen:combined
```
```
curl -X 'POST' \
'http://localhost:8080/conclude/' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-d '{
"input": "Extremely rich people around rookie who aren’t rich are likely targets, because frankly on a whim that extremely rich person can solve most of the poor persons material problems in an instant. That’s why rich people live in secluded communities so they don’t have to deal with the consequences of obscene wealth."
}'
```
Note to myself: maybe we should just create all summaries for the API (like an array context of `summaries`, each with a `source` and `value` (or similar, have to check for consistency with other things)). And then just select one of these for the web UI.Johannes Kieseljohannes.kiesel@uni-weimar.deJohannes Kieseljohannes.kiesel@uni-weimar.dehttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/94Use IBM Debater tools2021-03-24T13:59:58ZJohannes Kieseljohannes.kiesel@uni-weimar.deUse IBM Debater tools```
[13:20] Martin Potthast: Can we use some of the IBM
Debater tools to clean up / annotate the args.me
corpus? (also: @Sascha Bondarenko from #touché )
[13:28] Sascha Bondarenko: they provide a bunch of
different classifiers incl...```
[13:20] Martin Potthast: Can we use some of the IBM
Debater tools to clean up / annotate the args.me
corpus? (also: @Sascha Bondarenko from #touché )
[13:28] Sascha Bondarenko: they provide a bunch of
different classifiers including the one for the
argument quality, probably we can
```
@poma1006 @yemi8356
I did not have a look at the API. Still, I expect we can call it during assessment (like [snippet generation](https://git.webis.de/code-research/arguana/args/args-framework/-/blob/master/src/main/java/me/args/indexing/assessment/snippets/TextRankSnippetGenerator.java)). We should again use a [proxy cache](https://git.webis.de/code-generic/code-admin-knowledge-base/-/blob/master/services/args/snippets-args-me.k8s.yaml#L43) to make the repeated assessment of the same argument instantaneous.https://git.webis.de/code-research/arguana/args/args-framework/-/issues/92Show pro/con counts in UI2023-02-27T08:40:17ZJohannes Kieseljohannes.kiesel@uni-weimar.deShow pro/con counts in UIThe API exists now. We should query this API (including `contextId`) by JavaScript on loading a page. The API takes a bit as it has to get all arguments from the index, so we should directly start it and then replace loading indicators i...The API exists now. We should query this API (including `contextId`) by JavaScript on loading a page. The API takes a bit as it has to get all arguments from the index, so we should directly start it and then replace loading indicators in the menu.Github Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/91Update dataset on Zenodo2023-02-27T08:46:43ZJohannes Kieseljohannes.kiesel@uni-weimar.deUpdate dataset on ZenodoI guess we should wait for #89 but then make a new version on ZenodoI guess we should wait for #89 but then make a new version on ZenodoGithub Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/89Argument Snippet Generator Strange Cases2021-03-19T08:44:08ZJohannes Kieseljohannes.kiesel@uni-weimar.deArgument Snippet Generator Strange CasesPlease update the title once we can pinpoint the problem.
Here is a collection:
- https://dev.args.me/arguments/S47ca84a7-Ad12243ce?query=abortion
Selected "Thank you!" as title. [test.json](/uploads/68e02710ec534eea251d8c0546b4...Please update the title once we can pinpoint the problem.
Here is a collection:
- https://dev.args.me/arguments/S47ca84a7-Ad12243ce?query=abortion
Selected "Thank you!" as title. [test.json](/uploads/68e02710ec534eea251d8c0546b4b778/test.json) [test-single.json](/uploads/28a9c923f1dec813904de52bb18f78e6/test-single.json)
```
$ rid=$(hexdump -n 16 -e '4/4 "%08X" 1 "\n"' /dev/urandom) && curl -X POST "https://snippets.args.me/?id=$rid" --data @test.json -H "Content-type: application/json; charset=UTF-8"
[{"body": [{"span": [2133, 2318], "text": "Even if this speculative mechanism is true, it still would not be classified as an abortifacient because it would not expel fetal membranes or a placenta, as dictated in the definition."}, {"span": [6007, 6045], "text": "This has been a most enjoyable debate."}], "title": {"span": [6007, 6045], "text": "This has been a most enjoyable debate."}}, {"body": [{"span": [741, 778], "text": "It all eliminates the infant correct?"}, {"span": [2415, 2425], "text": "Thank you!"}], "title": {"span": [2415, 2425], "text": "Thank you!"}}, {"body": [{"span": [1872, 1942], "text": "I am against this, I don't believe I can make my argument any simpler."}, {"span": [6031, 6041], "text": "Thank you."}], "title": {"span": [6031, 6041], "text": "Thank you."}}, {"body": [{"span": [954, 1304], "text": "Immoral: -not in conformity with accepted principles of right and wrong behavior -wicked -not in conformity with the accepted standards of proper sexual behavior; unchaste; lewd (may I also note that many compare morality with Christianity as well as many other religious beliefs involving Christ) I care to dispute the fact that abortion is immoral."}, {"span": [3535, 3576], "text": "These also goes for all women in general."}], "title": {"span": [3535, 3576], "text": "These also goes for all women in general."}}, {"body": [{"span": [910, 1009], "text": "I start by arguing that there are many forms of abortion that should be prevented on moral grounds."}, {"span": [2115, 2125], "text": "Thank you."}], "title": {"span": [2115, 2125], "text": "Thank you."}}, {"body": [{"span": [58, 124], "text": "I will state my reasons in the second round in support of my side."}, {"span": [125, 220], "text": "If my opponent would be so kind to state his/her reasons in a brief statement in this round...."}], "title": {"span": [125, 220], "text": "If my opponent would be so kind to state his/her reasons in a brief statement in this round...."}}]
```
The argument in question is the second one, for which the following is returned:
```
{"body": [{"span": [741, 778], "text": "It all eliminates the infant correct?"}, {"span": [2415, 2425], "text": "Thank you!"}], "title": {"span": [2415, 2425], "text": "Thank you!"}}, {"body": [{"span": [1872, 1942], "text": "I am against this, I don't believe I can make my argument any simpler."}, {"span": [6031, 6041], "text": "Thank you."}], "title": {"span": [6031, 6041], "text": "Thank you."}}
```
However, if I send in just the argument, I get a better result:
```
$ rid=$(hexdump -n 16 -e '4/4 "%08X" 1 "\n"' /dev/urandom) && curl -X POST "https://snippets.args.me/?id=$rid" --data @test-single.json -H "Content-type: application/json; charset=UTF-8"
[{"body": [{"span": [63, 227], "text": "\"Whether or not abortion can also be classified as assault is irrelevant to this debate\" (I argue this statement) As a matter of fact it is relevant to this debate."}, {"span": [778, 926], "text": "Therefore, my opponent provided another form of abortion that he stated is the solution to pregnancy even though he is supposed to argue against it."}], "title": {"span": [63, 227], "text": "\"Whether or not abortion can also be classified as assault is irrelevant to this debate\" (I argue this statement) As a matter of fact it is relevant to this debate."}}]
```
- https://dev.args.me/arguments/Sb1869121-A8fd60f2c?query=abortion
No title
- https://dev.args.me/arguments/Sb186fda1-Ab9349bc5?query=abortion
URL as titleMilad AlshomaryMilad Alshomaryhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/76Get index term statistics2023-02-27T08:41:47ZJohannes Kieseljohannes.kiesel@uni-weimar.deGet index term statisticsOne can get the term statistics of an index in elasticsearch by sending a request like this: "index/_termvectors/id".
A Touche-participant asked for a similar API endpoint for args. This might be something that several API users could ...One can get the term statistics of an index in elasticsearch by sending a request like this: "index/_termvectors/id".
A Touche-participant asked for a similar API endpoint for args. This might be something that several API users could find useful.Github Transitionhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/67Loading indicators for aspect space2021-02-15T14:38:05ZJohannes Kieseljohannes.kiesel@uni-weimar.deLoading indicators for aspect spaceShow the nonagram at the start with a loading indicator at the center. Then add the dimension names once they are ready. Then add the points and remove the loading indicator.Show the nonagram at the start with a loading indicator at the center. Then add the dimension names once they are ready. Then add the points and remove the loading indicator.Dora KieselDora Kieselhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/63Archive all web pages that we crawl2020-05-26T19:32:22ZJohannes Kieseljohannes.kiesel@uni-weimar.deArchive all web pages that we crawlThe Internet Archive allows people to archive web pages. I'm not aware of the details of this process, but it seems we should go for an automatic archiving of the web pages using this service. This will allow us to add a link to the Wayb...The Internet Archive allows people to archive web pages. I'm not aware of the details of this process, but it seems we should go for an automatic archiving of the web pages using this service. This will allow us to add a link to the Wayback Machine to the arguments, which hopefully prevents situations like in #62.https://git.webis.de/code-research/arguana/args/args-framework/-/issues/61Decide which filter settings to keep when changing query and mode2020-03-02T10:41:12ZJohannes Kieseljohannes.kiesel@uni-weimar.deDecide which filter settings to keep when changing query and modeThe interface does not know which filter settings will still be available for a new mode or (in case of aspect filters) a new query.
Currently none is kept.
Probably it would be good to keep the mode and respective filter settings when...The interface does not know which filter settings will still be available for a new mode or (in case of aspect filters) a new query.
Currently none is kept.
Probably it would be good to keep the mode and respective filter settings when issuing a new query.https://git.webis.de/code-research/arguana/args/args-framework/-/issues/57Aspect space view: multiple selection2020-03-02T10:41:43ZJohannes Kieseljohannes.kiesel@uni-weimar.deAspect space view: multiple selectionIt is probably useful to select multiple aspects.
How should this be done interface-vice?
What does a multiple selection mean?
- Of of the selected has to be top-ranked?
- All of the selected have to be top-ranked? (thus have the s...It is probably useful to select multiple aspects.
How should this be done interface-vice?
What does a multiple selection mean?
- Of of the selected has to be top-ranked?
- All of the selected have to be top-ranked? (thus have the same weight)
- All of the selected have to have a top-n-rank (n being the number of selected aspects)?
The API supports all of these options, though maybe special considerations need to be taken for "Other": #56Dora KieselDora Kieselhttps://git.webis.de/code-research/arguana/args/args-framework/-/issues/56Aspect space view: select Other2021-04-30T21:41:34ZJohannes Kieseljohannes.kiesel@uni-weimar.deAspect space view: select OtherIt is currently not possible to select "Other".
This is due to technical limitations on the server side: "Other" is not known at indexing time. It would be possible to create a Lucene query that uses all aspects that are in "Other", but...It is currently not possible to select "Other".
This is due to technical limitations on the server side: "Other" is not known at indexing time. It would be possible to create a Lucene query that uses all aspects that are in "Other", but these are too many to fit into the maximum header size for GET queries.
Possible solutions:
- Use a "ranker" that interprets "Other" based on the aspectSpace parameter and filters out arguments. Pro: uses the information attached by the AspectSpaceProjector without the need to add more analysis. Con: very different to how filtering works for all other aspect ranks, thus requires an own API parameter.
- Make the Retriever interpret "Other" based on the aspectSpace parameter and split it into filters for each represented aspect. Pro: Makes "Other" behave as any other aspect from an API-point of view, with the exception that it requires the aspectSpace parameter. Con: probably less effective (but not sure about this), as this adds hundreds of clauses to the Lucene query.
I would prefer the second solution over adding a new API parameter, but that one may not be feasible (needs testing).