Computational Linguistics - Science topic

Hello, Arthur; is there anything particular in computational linguistics you are interested in?

هذا مرتبط بتخزين المفاهيم النفسية للكلمات، مثل: نار= عذاب. جنة=تواب، حجر=قسوة

وهكذا تتم المعالجة الآلية الإحصائية لينجح في نهاية المطاف الحاسب الآلي وفق خوارزمية متقنة في كشف التوجهات النفسية للنص المقروء .

Rafal Rzepka Hi there! This link is down, do you know how to access it now? Thank you so much!

Kuantum fiziğinde hesaplamalı dilbilimleri ve veri analizinde elde edilen bilgiler ile sonuca ulaşılır. Kuantum fiziğinin her durumuda bu geçerlidir. Bilgi ve iletişim araçlarında veya kendi alanında doğru olarak elde edilen verilerle hareket etmek ve bunu uygulamak en uygun metottur. Her zaman geçerli olan durumun uygulanması gerekir.

Kerry Sluchinski You might try the following user-friendly POS taggers/segmenters for Korean language data:

1. KoNLPy: KoNLPy is a Python module for Korean natural language processing. It features a POS tagger as well as numerous tools for Korean language processing. KoNLPy is straightforward and well-documented.

2. KOMORAN: KOMORAN is a Korean morphological analyzer and POS tagger that is free source. It is available as a command-line utility and as a Java library. For testing reasons, KOMORAN offers a user-friendly online interface.

3. Hannanum is a Korean morphological analyzer and POS tagger. It is a Java library that is built on a dictionary-based approach. Hannanum is simple to use and provides a user-friendly online interface for testing.

4. Korean Parser: Korean Parser is a dependency parser and part-of-speech tagger for Korean. It is written in Python and may be used as either a command-line utility or a Python library. Korean Parser is straightforward and well-documented.

5. Lingua-STS: Lingua-STS is a web-based tool for processing Korean language. It features a POS tagger as well as numerous tools for Korean language processing. Lingua-STS is simple to use and features an intuitive online interface.

These tools are all simple to use and may be used to separate Korean text data and conduct POS tagging.

Dear Dr Fatemeh Daneshfar . See the following useful RG link:

Dear Issa Annamoradnejad ,

Generally, academic institutes prefer Scopus/WOS indexed journals rather than conferences, and even give more credits for that.

Dear Matthew Abua Ebim

My comments are the same as those of Michal Místecký. If you need an overview of things, see if this can assist!

focuses in the domains

Hello Stefan Ziemendorff,

Praat could be practically useful for doing various tasks of phonetic analysis by computer.

More info:

Good luck,

أشكر لسعادتك الاهتمام والرد، وليتنا نتواصل بشأن هذا المجال فلست متخصصا فيه ولكن لدي بعض القراءات حوله، وهو أحد المحاور التي أتبناها في أبحاثي ومقالاتي حول الدراسات الثقافية والوعي العربي...

كل الشكر والتقدير لكم.

I would suggest using Standford CoreNLP to annotate your texts (corpus) with POS tags, and I believe this computational package can have different scripts for different languages. Then, extract the words with verb tags. Let me know if there is any question.

Definitely possible, but little bit complex...

I need it, too.

Pustejovosky's Generative Lexicon is also a model for lexical knowledge representation.

Standard datasets include WordSim-353 (see Agirre et al. 2009), SimLex-999 (Hill et al. 2016) and Chiarello et al. (1990). If you're interested in paraphrases including longer phrases, I'd also look at the Penn Paraphrase Database (PPDB, Ganitkevitch et al. 2013):

word error rate (WER%)

More recent POS taggers:

Farasa, ADAM, Multi-dialect Arabhic Pos tagging: A CRF approach and Part of Speech tagging for Arabic gulf Dialect using Bi-LSTM

Dear Gina,

Here you have a comprehensive list of Speech analysis and transcription tools for several languages.

I hope this helps.

Kind regards,

Begoña

You can try this free Arabic NER tool here:

Ian Maun, thank you very much! You helped me a lot!

Hi Trinadh

You might want to think about textual analysis tools. You could start by running a simple word frequency count to get an idea of the frequency of words within the reports. You could use the linux command line to get you started. Simple tools such as head, tail, wc, grep, tr, sort, sed, cut, uniq and awk can get you a long way in a very short space of time. You could also write some simple Python scripts to carry out these tasks. Python will run on most O/S, is simple to programme and very fast in operation.

However, at the high end of the spectrum, you could use a tool like GATE (written by the University of Sheffield) to perform full textual analysis of any kind of text. You can apply your own list of keywords to the text, or you can use the tool to develop a keyword list based on the contents of the files. It provides you with a POS tagger and many analysis tools which can allow you to understand the sentience of the content of the file. This is a serious tool, and not recommended for amateurs, as there is a steep learning curve. You will also need a thorough understanding of Natural Language Processing to get the best out of it. I have added a link for the GATE software to let you see what it can do.

I hope that helps.

Regards

Bob

Yes,

1. is the phrases in passive or active voice and what is proportion of both

2. What kind of pronouns are used in terms of level of politeness

3. What type of phrases: simple, subjectless, complex; with or without images;

4. Any proverbial references?

5. Plain or idiomatic what is the proportion of both

6. Then establish the features as a defining cluster for sub-genres in songs

thanks

You can add numpy.array objects by using the '+' Operator. You have to pay attention to the shape, but your arrays seem to have the shape (1,2).

Sou you can use the following code. This will return a np.array.

I am willing if you provide me with more info. 

See MADAMIRA :

The link of Malouf:

Interesting point regarding the CFA point above, quite a few linguistics papers that have done MDA via EFA have NOT then gone on to do CFA to confirm the model, e.g. Hardy & Romer (2013), http://uteroemer.weebly.com/uploads/5/5/7/7/5577406/hardy_and_roemer_2013_cor2e20132e0040.pdf.  

That said, I did a PCFA myself on the data, and got quite poor normed-fit and comparative fit indices, although RMSEA was appropriate.  It was interesting to attempt something like this, however, so thanks for the suggestion.

This is certainly not an exhaustive solution, but you can get a very substantial list of inflected forms and their lemmas from the Arabic Universal Dependency Treebank:

The Treebank contains about 242K lemmatized tokens, so I think those could be collapsed into a rudimentary lemma list covering quite a lot. The license is CC-BY-NC-SA, as shown here:

I think NTCIR 2 & 3 have a corpus like this (positive sentences):

You may check, also, Multiling2015 corpus (the summaries are manual)

MSS task:

MMS task:

MALL is usually defined as language learning that is supported through mobile devices or where the learner is mobile. It is not defined by the type of language-learning activity that learners perform through the mobile device/application. It's worth keeping in mind that anything that can be called MALL could also be called CALL since mobile devices are a type of computing device.

Here are some references which should give you a sense of the types of things that fall into the MALL category.

Kukulska-Hulme, A., & Shield, L. (2008). An Overview of Mobile Assisted Language Learning: from Content Delivery to Supported Collaboration and Interaction. ReCALL, 20(3), 271–289. http://doi.org/10.1017/S0958344008000335

Cui, Y., & Bull, S. (2005). Context and learner modelling for the mobile foreign language learner. System, 33(2), 353–367. http://doi.org/10.1016/j.system.2004.12.008

Hockly, N. (2013). Designer Learning: The Teacher as Designer of Mobile-based Classroom Learning Experiences (pp. 1–12). Monterey, CA, USA: The International Research Foundation for English Language Education (TIRF). Retrieved from http://www.tirfonline.org/english-in-the-workforce/mobile-assisted-language-learning/designer-learning-the-teacher-as-designer-of-mobile-based-classroom-learning-experiences/

Godwin-Jones, R. (2011). Emerging Technologies Mobile Apps for Language Learning. Language Learning & Technology (LLT), 15(2), 2–11.

Chinnery, G. M. (2006). Going to the MALL: Mobile Assisted Language Learning. Language Learning & Technology, 10(1), 9–16.

Palalas, A. (2011). Mobile-Assisted Language Learning: Designing for Your Students. In S. Thouësny & L. Bradley (Eds.), Second language teaching and learning with technology: views of emergent researchers (pp. 71–94). Dublin: Research-publishing.net.

Burston, J. (2014). The Reality of MALL: Still on the Fringes. CALICO Journal, 31(1), 103–125. http://doi.org/10.11139/cj.31.1.

Demmans Epp, C. (2015). Mobile adaptive communication support for vocabulary acquisition. Journal of Learning Analytics, 1(3), 173–175.

Demmans Epp, C. (2016). Supporting English Language Learners with an Adaptive Mobile Application (Doctoral). University of Toronto, Toronto, ON, Canada. Retrieved from http://hdl.handle.net/1807/71720

Please consider the datasets I mention in my PhD and related documents.

Dear Colleague,

The literature is vast. If you wish to focus on criticism, articles by Don Ringe (with various coauthors) may be useful. On a different note, the book "Historical Linguistics and Lexicostatistics"  ed. by V.V. Shevoroshkin and Paul Sidwell (Melbourne, 1999) could provide you with food for thought (especially the chapters by Sergei Starostin, Harald Sverdrup).

I attach a paper which contains relevant details of the maths applied in the revised version of glottochronology (my co-author, V. Blazek is the mathematician :-)).

Hope, this helps.

Best regards,

Irén H.

Hi Weichang,

If you're using a CQP based interface like CQPWeb for the BNC, you can use a part of speech regular expression to get a rough approximation of NPs. Using the CLAWS tags in the BNC, you could find two lexical NPs after a lexical non-gerund verb like this in CQP:

[pos="VV[^G]?"][pos="AT.*"]?([pos="R.*"]?[pos="J.*"])* [pos="N.*"][pos="(AT|J).*"] ([pos="R.*"]?[pos="J.*"])* [pos="N.*"]

The main trick is avoiding compounds of N+N, which can look like to consecutive NPs. Then you still have to filter out adverbial NPs, such as "eat [lunch] [every day]", but you can sample a subset to establish how frequent those are in your data. Either way, you won't get a perfect result, but a pretty good one.

If you have COCA in a CQP interface you could do something similar, depending on whether you're using COCA's tagset or re-tag it using some other tagset (Penn or CLAWS, for example).

Hope this helps,

Amir

Nowadays there are quite a big trend to research on deep learning of NLP and Sentiment Mining , you can follow research work of stanford to get a better understanding and continue your work.

Dear André,

do you remember some of the 2/3 of your thoughts which you thought in internal language? How do you do that?

Can you rethink them and discuss them internally? How do you do that?

Are among those thoughts some which cannot come into mind like a picture (because they are in some way more abstract)? How do you remember them, rethink them, and discuss them internally?

 Dear Hamzah,

In my knowledge, there is no such tool for this task in Arabic. I don't know if such tool does exist for other languages as well.

As you know there is no standard form of the nouns. You can at a first stage extract the nouns automatically. And select at the second stage the right nouns by implementing a  rule based system.   Arabic grammar experts can help you efficiently in this stage.

Is there any database or program which links arabic nouns to their derivatives (for example linking "اسخراج" to"استخرج")?

I remember there was one back several years ago. The database requires that the radicals of the basic measure verb be entered in a slot. The outcome includes the (required) derivation(s). I had the link and lost it. To answer your question, yes, there is, well, was. If you conduct a search, you may find it.

In a nutshell, LMS focuses on how a student learns while CMS focuses on how a course is delivered and implemented for learning.  In a nutshell...

You can use WEKA software. See the below link:

This presentation about Text Mining might be relevant for you:

I've started working on AntConc only recently; it helps you to view the files. However, it is not very good for extracting words or searching for collocations. Laurence Anthony (the designer) promised to develop a new version (4.0)  that can help you process Arabic texts without messing things up. Meanwhile, I'm working on Wordsmith, which is not free unfortunately. All the best!

Here is a list of relevant IR systems :

Stephen, that's really cool stuff! Thanks. 

There seems to be work by Sophie Scott.

I couldn't find a specific study, but

this seems to be related:

Other:

Regards,

Joachim

Usually sandhi is regarded as a generalized way to handle the joint between certain phonemes.  There may be many rules, but they all fall into the category of what happens when one word ends with one phoneme (in the case of tone, it seems to be an aspect of the ending phoneme) and the next begins with another.  So these examples just fall into cases handled by the system of sandhi. 

Hi, are you aware of the following open-source system?

I don't have any personal experience with the program but it seems it might worth a try. 

Hand over hand and side by side so they can see the perspective!

You need a tagged corpus for training your N-Grams and build a Language Model, and then use the Language Model  as a discriminative resource for word prediction.

You do not necessary need a treebank, in the sense that dependencies are not a must. What you need are Part-of-speech, lemma and morpho-synctactic information. With this info you can boost precision in prediction. We implemented this method for Italian, please see papers in my page.

For inflected languages, synctatic N-Grams can provide significative gain. In our experiments we reported  up to 30% improvement of KS (Keyword Saving).

Here is a recent paper.

I hope, it will help you.

Good luck

Greetings

I'm not sure what you mean by "innate". Do you mean "Do UG's claims actually match how people really process language?". To the best of my knowledge, there is no easy way to assess that: I'm not sure one could devise an experiment to validate the UG hypothesis as a whole. Maybe proving or disproving certain claims.

If you mean "Is Generative Grammar a consistent theory?", now that's another story. Even if one could prove it's consistent, it wouldn't necessarily mean it's true, i.e. it successfully accounts for language emergence, evolution, variation, acquisition, etc. in every aspect. If you look at other scientific fields, it seems to me that quantum mechanics AND relativity theory are both consistent models of some aspects of reality. Yet, they are fundamentally irreconcilable. Unless you start unifying them under an overarching model (string theory).

So it all boils down to personal belief, I think, hence the fierce arguments these questions generally spark among linguists.

This will depend on the type of features you will use for the  computational linguistic application. you should know if the used features are dependent or independent, linearly separable on non linearly separable etc. After specifying the nature of the used linguistic features, then you should search for the Neural Net that suits your training data.

thanks for your answers 

Well, yes. I know. Quite many people have tried to solve the problem but, to my knowledge, so far nobody has come up with a clear-cut model of the interface between natural language semantics and one or more kinds of logic. And I'm not quite happy with the metaphorical use of expressions like 'extract' and 'knowledge' in the context.

In the European project MANTRA we developed a parallel corpus based on EMEA and MedLine. It comprises of a silver standard large corpus (using different systems tagging the corpus with a majority vote) and a manually crafted 550 document gold corpus (only EMEA) used to calculate precision and recall of the silver standard corpus. The link for the downloads is https://sites.google.com/site/mantraeu/project-output and project information can be found at https://sites.google.com/site/mantraeu/

yeah, there is always need for such things. Possibility position can depend on the environment but the languages such written from right to left need such things in entirety.

For IE, in my knowledge, CRF is the best method for extracting the snippets of information. However, you need a training data?

Could you describe more detail what is your problem? Which is information you want to extract?

Our coding software (Profiler Plus) is commercial, but we have been working on a coding scheme to extract propositional data (such as causality) from text. It is still very much a work in progress. However, if you are interested in a collaborative project, I'm interested in working on concrete applications.

I didn't use textmining tools for Chinese language so far, but I find some information about two of them - maybe it will be useful:

The Stanford Word Segmenter supports Chinese:

There are some plugins to GATE for processing Chinese:

Hi Bram,

I'm an experienced Python user but don't really know much about that specific scientific discipline. In any case you have a Python library (Natural Language Toolkit) that should be able to help you a lot:

Besides this for numpy based data (arrays) you may want to look at (besides panda) scikit-learn for Machine Learning algorithms:

Matplotlib for plotting:

http://matplotlib.org/ (although there are plenty of others if you need more performance in plots)

I'm not sure what in what formats your raw data comes from but there's a lot of libraries for dealing with CSV, EXCEL formats, among others...

I would advise you to install everything with 64 bits version. Python has official 64 bit releases but some libraries don't so this link might help you in that regard:

Besides I would advise you to install a more complete package set instead one library at a time. WinPython (for windows) has an excellent installation manager (besides already coming with a lot of stuff) :

Anaconda is another distribution (64 bits) also good for Mac or Linux:

Hope it helps.

There is some work on taxonomy extraction from Wikipedia (identifying taxonomic links in the Wikipedia catetory graph).

You might want to check our paper

García, Renato Domínguez, et al. "Automatic taxonomy extraction in different languages using wikipedia and minimal language-specific information." Computational Linguistics and Intelligent Text Processing. Springer Berlin Heidelberg, 2012. 42-53.

see http://link.springer.com/chapter/10.1007/978-3-642-28604-9_4

or ftp://ftp.kom.tu-darmstadt.de/papers/DSRS12-1.pdf

Hi! This is not my field, but it seems to me that this ERP study uses pseudo-words as prime and figures as target and finds also a N400 component: Kovic, V., Plunkett, K. and Westermann, G. (2010). The shape of words in the brain. Cognition 114, 19–28.

You are probably looking for similarities among languages that may be related. Then, compile a list of common words for each language: mother, father, brother, sister, man, woman, child, house, rice, sun, to go, etc

You can look at Ellen Riloff's work :

In there find: NaturalJava: A Natural Language Interface for Programming in Java.

While it is dated you can use it and Joachim's reference to get a head start and maybe find some other references..

Maybe Ken Hyland's  stuff on Academic Discourse ,namely his work on certain verbs  in research articles.

For HMM algorithm, you can use HTK (HMM tool Kit). HTK can do automatic segmentation and labeling.

Hi Ibrahim,

I don't believe so. Please check out the following paper for the issues involved in developing such corpora: http://archimedes.fas.harvard.edu/mdh/arabic/NAACL.pdf.

Hi,

There is a work of Meselhi https://www.researchgate.net/publication/268513573_Novel_Hybrid_Approach_to_Arabic_Named_Entity_Recognition

All the best,

Sebastian, your a talking about a different problem. The task of defining the tag set is vitally important and usually done badly. Multiple experts in creating those definitions are valuable, but they don't all have to do the annotating. In fact in one of our studies  we showed that the experts were more constant between themselves but missed more of the peripheral examples expressed in non-standard language, higher precision, but lower recall. We use linguists for doing annotations as they are much more aware of the nuances of language and more conscientious about the accuracy of the tagging. I would always advise  to not use domain experts to do the annotating as they don't have the language analysis tools. Their role is to specify the task the tag definitions and arbitrate difficult cases, however it is the linguists who refine the tag definitions after the experts have created them.

Dear Ahmed

the problem with left recursion, from a computational linguistics point of view, is that it leads to infinite recursion, as mentioned in other posts. And, sadly, linguists do tend to write an awful lot of such rules, as the example below shows (a very naive DCG grammar for English relative clauses). If you 'consult' this grammar with swi-prolog, all will apparently run smooth because swi-prolog can deal with such recursive rules appropriately. If you submit the following  goal "s([the,man,that,he,knows,sleeps],[]).", you'll get "true" as an answer. But, if you ask swi-prolog to search for more results (";"), then you'll get an "Out of local stack" error because of the left-recursion. 

The general strategy is "transform your left-recursive rules into right-recursive ones". It means you must tweak your grammar to eliminate such left-recursive rules and transform them into right-recursive ones, with the help of an intermediate non-terminal (cf. for eg. http://web.cs.wpi.edu/~kal/PLT/PLT4.1.2.html). 

From an algorithmic point of view, different approaches have been published, in order to deal with such left-recursive rules (as said earlier, this is how linguists spontaneously write formal grammars). If you're looking for algorithms, you can have a look at Bob Moore's paper http://research.microsoft.com/pubs/68869/naacl2k-proc-rev.pdf.

s --> np, vp.

np --> det, n.

np --> np, relc.%this is a left-recursive rule

relc --> pror, pro, vt.

vp --> vi.

vp --> vt, np.

det --> [the].

n --> [man].

pro --> [he].

pror --> [that].

vt --> [knows].

vi --> [sleeps].

The methodology for comparing n machine learning (ML) methods over m data sets is described in (Demšar, 2006). Please see the link.

I show you an example in the attached document.

Apart from NLTK and Pattern, the only other Python NLP library that I've come across is TextBlob ( http://textblob.readthedocs.org/en/dev/ ):

"TextBlob is a Python (2 and 3) library for processing textual data. It provides a simple API for diving into common natural language processing (NLP) tasks such as part-of-speech tagging, noun phrase extraction, sentiment analysis, classification, translation, and more."

I've not used it extensively, so I can't comment on whether or not it would be a good alternative to pure NLTK (TextBlob is apparently built on top of both NLTK and Pattern). It does seem to attempt to simplify the code for certain aspects of text processing such as part-of-speech tagging, so it may suit your teaching needs as an introductory tool.

It is free, only that I don't know how versatile it is and what are other alternatives.

I have studied the relationship between the pitch and energy contour in my PhD  and summarized some of the work in the follow publication. Whilst it is not directly related to tonal languages there is a comprehensive analysis and discussion about the  relationship between the two measurements. 

The Relationship between Prosody and Breathing in Spontaneous Discourse

Hird, Kathryn ; Kirsner, Kim

Brain and Language, 2002, Vol.80(3), pp.536-555 [Peer Reviewed Journal]

Digital Thesis available from UWA

Are cognitive processes involved in the control of declination in spontaneous speech?

Kathryn Marie Hird University of Western Australia. Dept. of Psychology. 1994

 

Available at Reid Library  Store  (Thesis 1994 HIR)

The group of Martin Vechev is using language models (and other NLP techniques) in program analysis and synthesis, e.g.:

Hello,

Just Try to make a combination between HMM and GMM. 

 Hi, Sentiment analysis itself faces some challenges from technical perspective in any context not just the smart cities. The general challenges will be domain-dependency, context-dependency, sarcasm, spam, ambivalency, and implicit opinions(or sentiment). These are the challenges that have to be considered in the design and implementation of any application of sentiment analysis.

Sir, You can write to John M Swales who was instrumental in developing MICASE. He responds to our queries. Generally we get access to transcripts only. The audio databases are not shared. There is Dr Claudia from Germany, Dresden . she collected a lot of samples from Indian users of English.  Her contact is also useful.