[Corpora-List] methods to identify the direction of causation

Marco Baroni marco.baroni at unitn.it
Thu Jan 5 19:59:22 CET 2017


Thanks for the further advice.

To clarify, what I'm looking for is papers that look specifically at the relation between out-of-context "types", not sentence-based "instances".

For example, Mirza and Tonelli paper classify "compliance"-"inspections" as a causal pair, because in a particular sentence *complying* to some regulations caused the *inspections*.

While this is arguably the more interesting case (finding what causes what in a specific situation), we are right now working on a model that only looks at prototypical relations between concepts, independently from specific situations, as in the "virus"/"death" example, where, if you ask subjects out-of-context, they will tell you that it's the virus that causes death and not vice versa, even if you can think of specific situations where, say, it's a death that caused the spreading of a virus, etc.

Thanks again.

Best,

Marco

On Thu, Jan 5, 2017 at 7:14 PM, Ken Litkowski <ken at clres.com> wrote:


> In addition to these good references, there's a paper from COLING 2016, CATENA:
> Causal and Temporal Relation Extraction from Natural Language Texts
> <https://www.aclweb.org/anthology/C/C16/C16-1007.pdf>, by Paramita Mirza
> and Sara Tonelli. This is interesting to me since it makes use of PDEP
> <http://www.clres.com/db/TPPEditor.html> preposition data to facilitate
> this extraction. That is, prepositions can be of some help in identifying
> the direction of causation ("death from virus").
>
> On 1/5/2017 11:21 AM, Francis Bond wrote:
>
> There is a lot of recent work on a very large scale by NICT in Japan.
>
> Here is a recent summary:
>
> WISDOM X, DISAANA and D-SUMM: Large-scale NLP Systems for Analyzing
> Textual Big Data, Junta Mizuno, Masahiro Tanaka, Kiyonori Ohtake, Jong-Hoon
> Oh, Julien Kloetzer, Chikara Hashimoto and Kentaro Torisawa, In the
> Proceedings of the 26th International Conference on Computational
> Linguistics (COLING 2016) (Demo Track), Osaka, Japan, December, 2016.
>
> and there is a lot more here:
>
> http://www2.nict.go.jp/direct/publications-e.html
>
> On Thu, Jan 5, 2017 at 8:04 AM, Xu, Jiajin <ustcxujj at gmail.com> wrote:
>
>> Dear Marco,
>>
>> I just fished out the following references on my hard drive, which I hope
>> will be of some interest to you. They might be somewhat dated though. Also,
>> please excuse me for not editing the references in any standard
>> bibliographical style.
>>
>>
>> 1. Text Mining for Causal Relations Roxana Girju and Dan Moldovan
>> <http://dl.acm.org/citation.cfm?id=708596>
>> 2. Blanco, E., N. Castell, D. Moldovan. 2008. Causal relation
>> extraction.
>> 3. Garcia, D. 1997. COATIS, an NLP system to locate expressions of
>> actions connected by causality links.
>> 4. Girju, R. 2003. Automatic detection of causal relations for
>> question answering.
>> 5. McNorgan, C., et al. 2007. Feature-feature causal relations and
>> statistical co-occurrences in object concepts.
>> 6. Suppes, P. 1970. A Probabilistic Theory of Causality. Amsterdam:
>> North-Holland Publishing Company.
>> 7. Swanson, D. 1991. Migraine and magnesium: Eleven neglected
>> connections.
>>
>> Best,
>>
>> Jiajin XU
>> Ph.D., Professor
>> National Research Centre for Foreign Language Education
>> Beijing Foreign Studies University
>> Beijing 100089
>> China
>> http://www.bfsu-corpus.org
>>
>> On Thu, Jan 5, 2017 at 8:23 PM, Marco Baroni <marco.baroni at unitn.it>
>> wrote:
>>
>>> Dear all,
>>>
>>> With a few colleagues, we're working at the problem of identifying the
>>> prototypical relation of causation between words/concepts (e.g., virus
>>> causes death, rather than vice-versa).
>>>
>>> We are framing this as an out-of-context, concept-level task (that is,
>>> we're not trying to identify the cause relation as it is expressed in
>>> specific sentences, but as it holds canonically between word pairs).
>>>
>>> This is a new area for me, so I'd be grateful for pointers to papers
>>> that have proposed corpus-based methods to address this task.
>>>
>>> Thanks in advance,
>>>
>>> Marco
>>>
>>>
>>> _______________________________________________
>>> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
>>> Corpora mailing list
>>> Corpora at uib.no
>>> http://mailman.uib.no/listinfo/corpora
>>>
>>>
>>
>> _______________________________________________
>> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
>> Corpora mailing list
>> Corpora at uib.no
>> http://mailman.uib.no/listinfo/corpora
>>
>>
>
>
> --
> Francis Bond <http://www3.ntu.edu.sg/home/fcbond/>
> Division of Linguistics and Multilingual Studies
> Nanyang Technological University
>
>
> _______________________________________________
> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
> Corpora mailing listCorpora at uib.nohttp://mailman.uib.no/listinfo/corpora
>
>
> --
> Ken Litkowski TEL.: 301-482-0237 <(301)%20482-0237>
> CL Research EMAIL: ken at clres.com
> 9208 Gue Road Home Page: http://www.clres.com
> Damascus, MD 20872-1025 USA Blog: http://www.clres.com/blog
>
>
> _______________________________________________
> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
> Corpora mailing list
> Corpora at uib.no
> http://mailman.uib.no/listinfo/corpora
>
>

-- Marco Baroni Center for Mind/Brain Sciences (CIMeC) University of Trento http://clic.cimec.unitn.it/marco -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 11774 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20170105/56788674/attachment.txt>



More information about the Corpora mailing list