In general, we are interested in whatever helps, will help eventually, or might help computers meaningfully deal with language data.

The conference is intended to encourage exchange of opinions between the scientists working in different areas of the growing field of computational linguistics and intelligent text processing. Our idea is to get a general view of the state of art in computational linguistics and its applications.

Areas of interest include, but are not limited to, the following topics, provided that the work is presented in computer-related or formal description aspects:

We welcome works on processing any language (not necessarily English), though major languages are of more general interest. When discussing phenomena of languages other than English, please keep your discussion understandable for people not familiar with this language.

You can have a look at the contents of the proceedings of past CICLing events to get an idea of our interests. If not sure whether your topic is of interest, please ask us.

Traditionally, our keynote speakers give a formal talk, which is also published in extenso in the proceedings, and also organize a "special event" (a discussion, tutorial, experiment, or something just interesting). Such events, as well as publication of the keynote talks in the proceedings, are distinctive features of CICLing. [Past participants' opinions]

Jun'ichi Tsujii		Jun'ichi Tsujii Professor, Department of Computer Science, University of Tokyo, Japan; Professor of Text Mining, School of Computer Science, University of Manchester, UK; Scientific Director, National Center for Text Mining (NacTeM), Manchester, UK  NEW  Monday Feb 21 Computational Linguistics and Natural Language Processing Researches in Computational Linguistics (CL) and Natural Language Processing (NLP) have been increasingly dissociated from each other. Empirical techniques in NLP show good performances in some tasks when large amount of data (with annotation) are available. However, in order for these techniques to be adapted easily to new text types or domains, or for similar techniques to be applied to more complex tasks such as text entailment than POS taggers, parsers, etc., rational understanding of language is required. Engineering techniques have to be underpinned by scientific understanding. In this paper, taking grammar in CL and parsing in NLP as an example, we will discuss how to re-integrate these two research disciplines. Research results of our group on parsing are presented to show how grammar in CL is used as the backbone of a parser. A full-text paper on the topic is included in the proceedings.
Hans Uszkoreit		Hans Uszkoreit Professor of Computational Linguistics at the Dept. of Computational Linguistics and Phonetics of Saarland University at Saarbrücken. Scientific Director at the German Research Center for Artificial Intelligence (DFKI) and Head of DFKI Language Technology Lab., Germany Tuesday Feb 22 Formal talk (morning): Learning Relation Extraction Grammars with Minimal Human Intervention: Strategy, Results, Insights and Plans Systems that recognize instances of n-ary relations in large volumes of text have numerous practical applications. One of the most intriguing but at the same time most challenging approaches to complex information extraction is the bootstrapping paradigm that starts from a very small set of semantic examples, called the seed, for discovering patterns or rules, which in turn are employed for finding additional instances of the targeted information type. These new instances will then be used as examples for the next round of finding linguistic patterns and the game repeats until no more instances can be detected. Since the seed can be rather small, containing between one and a handful of examples, this training scheme is usually called minimally supervised learning. After a few cycles such systems may find thousands of instances.  As we could demonstrate, a truly successful application of the paradigm requires learning data that exhibit certain graph-theoretical properties. We developed a special approach to minimally supervised learning of relation extraction grammars that incorporates more linguistic processing than its predecessors. With this system we explored the potentials and limitations of the paradigm. From the gained insights we derived a medium-term research strategy. Part of this strategy are chains of experiments with fine-grained diagnostics. The results of these experiments have determined the evolution of the approach. Our current system works learns from positive and negative evidence using confidence estimation for the filtering of learned rules and instances. In my talk I will pursue four goals: provide an overview of the approach and its evolution,   explain and illustrate the adopted research strategy,   summarize the obtained results and their relevance for the strategy,   outline planned steps and open problems. The talk presents joint work with Feiyu Xu.  NEW  Special event (evening): Language Technology 2020 - Hunches, Hopes and Visions: “The closest things to crazy are growing close to you...” In Europe, we are currently in the process of formulating visions on LT applications, technologies and research paradigms in ten years. This process is organized by the European Network of Excellence META-NET, which I am coordinating. The nodes of this network are 44 European research centers in 31 countries. One objective is a Strategic Research Agenda for European Language Technology, especially addressing the problem of technologies for the multilingual European information society. This Agenda will hopefully influence the upcoming 8th Framework Programme (2014–2020) which will be prepared during the next two years. Since it may also have an impact on national research planning we are closely working with representatives of EU member states and language communities. We have established three Vision Groups that prepared interesting ideas and reports. In my special event, I would like to first share some of the ideas of the three groups and of distinguished colleagues outside the groups with the CICLing participants and then pick their brains for reactions and additional ideas. One historical observation should help us to shape this evening: Many courageous visions that later became validated by reality had been considered somewhat crazy at the time they were first conceived. Powerful phantasies usually do not emerge on planning boards. So let us relax from our latest gains in precision, recall, accuracy and run-time efficiency for some hours and open our minds and mouths in less controlled manners for an exercise of collective late-day dreaming.
Diana McCarthy		Diana McCarthy Lexical Computing Ltd; Erasmus Mundus Visting Scholar at Saarland University: Computational Linguistics and Phonetics Group and University of Melbourne Language Technology Group  NEW  Thursday Feb 24 Measuring Similarity of Word Meaning in Context with Lexical Substitutes and Translations Representation of word meaning has been a topic of considerable debate within the field of computational linguistics, and particularly in the subfield of word sense disambiguation. While word senses enumerated in manually produced inventories have been very useful as a start point to research, we know that the inventory should be selected for the purposes of the application. Unfortunately we have no clear understanding of how to determine the appropriateness of an inventory for monolingual applications, or when the target language is unknown in cross-lingual applications. In this talk I will examine various datasets which have alternative annotations of lexical meaning on the same underlying corpus data. These datasets are: The English Lexical Substitution dataset which contains lexical paraphrases of words in context (produced in collaboration with Roberto Navigli) The Cross Lingual Lexical Substitution dataset which contains lexical translations of words in context (produced in collaboration with Rada Mihalcea and Ravi Sinha) Two Usage Similarity Datasets which contains similarity judgments of a target word in a pair of contexts (produced in collaboration with Katrin Erk and Nick Gaylord) I will begin the talk by describing these datasets and the rationales behind their creation. I will then present new analysis to demonstrate that an overlap in lexical paraphrases (substitutes) between two uses of the same lemma correlates with an overlap in translations. I will also compare the degree of overlap in paraphrases and translations with annotations of usage similarity on the same data. The correlation bodes well for using any of these methods to evaluate unsupervised representations of lexical semantics. I will highlight that the relationship breaks down for some lemmas that but this behaviour on a lemma by lemma basis itself correlates with low inter-tagger agreement and higher proportions of mid-range points on the usage similarity datasets. Lemmas which have many inter-related usages might potentially be predicted from such data.
Christopher Manning		Christopher Manning Associate Professor of Computer Science and Linguistics, Sony Faculty Scholar, Natural Language Processing Group, Stanford University  NEW  Friday Feb 25 Holistic models of linguistic structure: Discriminative joint models and domain adaptation The main mass natural language applications for regular people are high-level, semantically oriented ones which have to just work in diverse contexts: question answering, machine translation, machine reading, speech dialog interfaces for robots and machines, etc. Humans are very good at these types of tasks, in part because they naturally employ holistic language processing. They effortlessly keep track of many layers of low-level information, while simultaneously adapting across topics and accents and integrating in long distance information from elsewhere in the conversation or document. In contrast, much NLP research not only focuses on lower-level tasks, like parsing, named entity recognition, and part-of-speech tagging, but, for the sake of efficiency, the models of these phenomena make extremely strong independence assumptions, which completely decouple these tasks. Models commonly only look at local context when making decisions and have no models or only very crude models for domain adaptation. I will present some of our recent work that begins to address these problems. I present a discriminative parsing model (a CRF-CFG) which can incorporate many source of evidence in decision making, and then show how that can be extended into a joint model of parsing and named entity recognition. This joint model improves the performance of both independent NER and parsing models, showing that joint information effectively flows between the tasks, and also supports the recognition of hierarchical named entities. This basic model does not address domain adaptation, and a practical complication of training such joint models is that they require jointly annotated data, and there is much less jointly annotated data available than data annotated with just parse trees or just named entities. In the last part of the talk I will describe the use of hierarchical models which allow us to address both these problems. I show that we can substantially improve the performance of our joint model by using domain-specific and non-jointly annotated data. This talk presents joint work with Jenny Finkel.

[Passed] Paper submission is divided in two phases: first, the authors should express their interest by registering a tentative abstract of the paper. At this stage, the full text is not yet required; the tentative abstract will only be used to find suitable reviewers for your paper. The authors of registered papers should upload the full text of the paper by the corresponding deadline.

Though we cannot guarantee processing of any paper that does not arrive by the corresponding deadline, you may contact us if you really cannot submit your paper in time, and we will see what can be done.

 NEW:  There will be two time slots for on-site registration:

• Feb 19 Sat, 19:00-20:00 lobby of official hotel.
• Feb 20 Sun, 19:00-20:00 lobby of official hotel.

Those who go to Kamakura must register on Feb 19; if you cannot, contact Alexander in advance. We strongly advise that you register either on 19th or 20th; if you cannot, we will try to handle your registration during the conference, but this is highly discouraged.

We did our best to keep the fee low, and we obtained some sponsorship from a private fund. However, Japan is extremely expensive, so we could not keep the fee as low as we wish it were, though much lower than we calculated originally. Registration fee is:

• Full: 45,000 Japanese yen (approx. US$ 540,  € 420)
• Student: 30,000 Japanese yen (approx. US$ 360, € 280)

The fee is per participant and per paper. I.e., for each paper, at least one fee is to be paid: if a participant presents two papers, two fees are to be paid. On the other hand, a participant that does not present any paper, should pay a fee.

You can pay on site, in cash; another way to pay is via bank transfer if you wish (please contact us for payment instructions via bank transfer). Please calculate the amount and sign your promise to pay by filling in the registration form. Some people have contacted us for a reduced fee. We will be able to provide a very limited number of reduced registrations or waivers. We will contact those people directly (please contact us if you really cannot pay the fee).

Recall that by submitting a paper, at least one author thereby have promised, in case of acceptance, to attend the conference in person to present the paper and to pay the corresponding registration fee. Also, please check our video recording legal notice.

On reduced registration fee: A very limited number of reduced registrations may be available. To apply, please contact us and thoroughly justify your application. Eligible for reduced registration are people from underdeveloped countries in case if their institutions have real difficulties paying the full fee (generally not included: North America, Western Europe, China, South Korea, Japan, but if you feel your situation is really different, try applying anyway). Authors must apply for reduced registration (clearly indicating the discount amount) before submission of their paper, and also must tick the group "[X] Discount or waiver is requested" in the web submission system. No applications will be considered for already reviewed papers. Notes: (1) Though all papers are judged by strictly academic criteria, for borderline cases and between papers of comparable quality we may give preference to papers with fully paid fee. (2) Though we will do our best for this not to happen, we cannot guarantee providing the material (including the proceedings) to participants with reduced fee. Also, in case of lack of seats in the excursion bus we will have to give preference to fully registered participants.

We are interested in a few  volunteer helpers , please contact us. Japanese speakers are preferred.

All papers accepted for oral presentation will be published in a proceedings volume edited by Springer in its Lecture Notes in Computer Science series, which is indexed in many major indices. Papers accepted for poster presentation will be published separately in a special issue of a journal to be announced later, see Poster Session (let us know if you actually prefer the poster type of publication).

In addition to the text of the paper, authors are strongly encouraged to provide programs that permit to reproduce their results, see CICLing verifiability, reproducibility, and working description policy.

Before submitting, please check our video recording legal notice. Submissions are received electronically.

Contact: See email options, fax, and the street address on www.CICLing.org/contact.html. Please avoid sending us any physical mail: we strongly prefer electronic communication.

Starting from this year, CICLing start implementing a new policy of giving preference to papers with verifiable and reproducible results:

   If the authors claim to have obtained some results, we encourage them to make all the input data necessary to verify and reproduce the results available to the community.

   If the authors claim to advance human knowledge by introducing an algorithm, we encourage them to make the algorithm itself -- and not only its (usually vague and incomplete) description -- available to the public.

We do not refer here to any demo or tool based on your paper, but instead to a form of proof and working description of the algorithm in addition to the verbal description given in your paper. An approximation of the idea is the code submitted with Church & Umemura's paper to be permanently hosted at CICLing servers, and cited in the paper (see last line): you see, we don't mean anything complicated. Obviously, you are also encouraged to show demo programs or tools based on your method, either as part of your talk or (better) at the demo session, and we will also be happy to host on our servers such software that complements your paper. However, this is not required. In contrast, we do believe that a publicly published scientific paper must be accompanied by a minimal working description of the algorithm, open-source and available to the community.

We do not yet have specific rules: we hope to elaborate the rules basing on this year's experience, so please use common sense. See the problems this policy is to address, as well as the list of software reviewing committee and instructions for the reviewers.

We expect to give a special best verifiability, reproducibility, and working description award to the authors of the software that in the best way fulfills the above goals (that is: the simplest and clearest code that proves the claims of the paper and allows one to exactly reproduce them).

Therefore, we encourage the authors to submit, together with the paper, a program (open source), as simple as possible, that follows the described algorithm and generates the results presented in the paper. We do not need any sophisticated interface or performance improvements, only easy to understand source code that generates the claimed results. Think of it as a proof of a theorem: the result reported in your paper is a theorem, and the source code generating this result is its proof. Our sole purpose is to exactly reproduce your results and to be sure that it is reproduced with exactly the same method as you describe. Minimalistic approach is the best: just implement your algorithm in a way simple to read and understand, nothing else. Please extensively comment your code. Naturally, input data are to be presented together with the code, otherwise how would we generate the same result? When this is impossible, you can provide instructions on where the data can be obtained (e.g.: WordNet, Google pentagrams, etc. need not to be included with your code, but we do need instructions on how we can obtain exactly the version you used; using standard software or corpora is highly preferable over home-made ones when possible).

CICLing will keep the right, though no obligation, to host your files on its servers. Upon acceptance of your paper, we will give you a permanent link to the hosted data; please indicate this link in the camera-ready version of your paper. Please accompany your code with a suitable license that would allow its free distribution, free study (and reverse-engineering if needed) by the public, and free use of the knowledge obtained from such a study. For the future editions of CICLing we plan to elaborate a special CICLing license for academic code, documentation, and data; if you have any ideas or suggestions on such a document, please let us know, any guidance is highly appreciated.

Submission of such code is not a requirement. For example, the nature of the paper may not require any additional data or code, or your experimental setting does not allow it, even after you have done all reasonable effort to make it possible. However, if the reviewers judge that the paper does require and does allow submission of the code and data to be verifiable and reproducible, then preference will be given to papers accompanied by the code. We do understand that you may not have had time to prepare the code. Again, we will use common sense in applying this policy.

Having said this, we recommend that you submit your code as a ZIP file attachment (in EasyChair, use the Attachment field on the paper submission page; if you didn't have time, try sendng us the ZIP file by email later) containing in its root the following directories (you may choose another structure if it makes more sense):

   File README.*, in appropriate format, such as PDF or TXT, with complete and clear description of the contents and its use, including installation, compilation, running, analyzing the results, and matching the results to the claims of the paper. For example, if Table 1 claims that your algorithm gives 74% on your corpus, describe where this figure is in the output of your program. Please specify versions of common software (such as Perl) on which you tested your program. Instead of including all material, the README file can point to other files, such as INSTALL.PDF.

   File LICENSE.*, in an appropriate format, specifying the license terms for your software, compatible with free distribution, studying, and reverse-engineering of your code. Anyone should be permitted to modify or use your software for any purpose. Wikipedia's approach to licensing is a good example of what science (i.e., advance of human knowledge) is meant to be. It would be good if this permission included the data (and not only software), but this is not always feasible; use common sense. Please specify who the authors are and how to contact them (to reduce spam I'd not recommend including emails; consider pointing to their webpages). The authors retain all author rights, even if they grant non-exclusive distribution permission to CICLing. Note: for the review stage you can omit the authors' names and contacts, but please don't forget to contact us for an update upon acceptance: it is very important that the user know it's your work.

   File LICENSE_CICLING.*, in an appropriate graphical format, should be a scanned image of a written document signed by at least one of the authors, which explicitly authorizes CICLing to host all the other files on its servers for unlimited time and to provide public access to them. (If you prefer so, you can provide this file upon acceptance of the paper.)

   Directory SOURCE with the source code of a program that implements your algorithm and produces the results you report. If possible use well-known major programming languages.

   Directory BIN with compiled executables for a major OS (Windows, Linux, or Mac).

   Directory UTILITIES with auxiliary programs that are not part of your algorithm. For example, if you use a grep program, you can put it here. Less common utilities must be included, along with their respective documentation.

   Directory INPUT with the input data necessary to produce the results, including corpora, dictionaries, grammars, or whatever may be needed.

   Directory OUTPUT with exactly the same output data that is expected to be produced if one follows your instructions. We will know that the program installed and ran correctly if by running your program we obtain exactly the same results.

   File RUN.*, such as RUN.BAT, that can be called without any parameters and the data from INPUT into the result in OUTPUT without any user intervention. The instructions should specify what file is to be called, possibly after some configuration.

   Other directories as appropriate, with meaningful names.

Please keep things as simple as possible (though not simpler), and the installation and use instructions as clear as possible. Usually this implies detailed and specific instructions, as well as completely automatic script that performs all necessary steps without user intervention. In the text of your paper, make sure that the software reviewers will easily locate the description of the input data and the obtained results; section titles such as Main Algorithm, Experimental Methodology, and Experimental Results could be helpful. If the reviewers fail to quickly and easily install your program, run it, and interpret the results, they will probably give up and lower your score.

The archive should be as self-contained as possible: it is a good idea to include as much as possible -- compilers, interpreters, utilities used, etc. If possible, to include a complete distribution of Perl or WordNet is better than to rely on that we will somehow find the version you used. Remember that science is done for eternity: your results should be reproducible in twenty years, when it may be impossible to find a specific version of a utility or compiler you used. If the resulting file is too large so that the system does not allow uploading it, please submit your paper alone and contact us for the attachment.

No double-blind policy for software: For the time being, we encourage the software to be anonymous but we do not require this. We understand that in some cases it can be impossible or too labor-consuming. If you really cannot make your software anonymous, then leave it as is.

The following awards will be given:

 Best Paper awards have been assigned by the Award Committee basing on the reviewers' scores and judgment of the Committee members. The criteria taken into account are: novelty, originality, and importance of the reported work and overall quality of the paper.

Winners:

• First place: Hiram Calvo, Kentaro Inui, and Yuji Matsumoto. Co-related Verb Argument Selectional Preferences.
• Second place: Shoji Fujiwara and Satoshi Sekine. Self-Adjusting Bootstrapping.
• Third place: Daniel Andrade, Takuya Matsuzaki, and Jun’ichi Tsujii. Use of Dependency Structure for Bilingual Lexicon Creation.

The papers accepted for the poster/demo session are anticipated to be published in a special issue of a journal, to be announced later. See here the guidelines for submitting and preparing your poster.

Poster session will be combined with the welcome party, so people will be in good mood when reading your poster. In our experience the authors often have better opportunity to communicate their idea to interested attendees via individual live interaction at a poster presentation than via standard talk. If you feel your paper is not competitive enough for the oral session, do go ahead and submit it: a poster can be an excellent opportunity for you to get feedback.

Before the poster session, on Monday afternoon, the poster papers will be presented (or we can call it announced) orally. Each presentation will be of one minute, and you can use a couple of slides. It is probably a good idea to use one or two slides to show the title and the main idea of your work, and another slide to briefly show an image of your poster, for people to recognize your poster during the poster session. The purpose of this one-minute presentation is not to explain your work in detail but to attract attention of people to you and your poster. If you succeed, you will then have two hours to explain your work to all interested people, during the poster session and in fact all the breaks on other days. Thus, please only include the main "selling points" of your work in your short presentation. Too much information would only confuse people.

 NEW  To streamline the session, we have asked you to send us your presentation, and we will pre-load it to a laptop. If you have your laptop at the conference, then please also have the same presentation prepared on your laptop in case if for any reason it does not run correctly on ours.

The poster session will be on Monday afternoon. It will be in this hall, the same place where we will have lunch (tables will be partly removed for the poster session) and not the same place where the talks will be.

There will be no Internet access. There may not be enough electricity outlets or power cables, so if you plan to use your laptop, be sure to have enough battery power. If you still plan to use an outlet, please bring your own T-joint, for other people to be able to use the same outlet.

The poster panels for the posters will be approximately 180 cm high x 120 cm wide -- this should be enough even for a horizontal A0 sheet, though a vertical A0 design might be safer. The posters will be pinned to the poster panels. See photo of the poster panels, another photo. We suggest that you include your photo on your poster, near the title: while you are looking at other posters, people interested in your poster would find you.

After the poster session on Monday afternoon, we suggest that you display the same poster at the conference (where the talks will be). On Monday, after the poster session you will have to return your poster back to the hotel. On Tuesday, you can bring it to the conference venue (which is different from the poster session venue) and leave it there displayed until the end of the conference, for people to be able to read it during breaks. We will provide you with poster panels at the conference place, to put your poster on Tuesday to Friday. Your photo on the poster is especially important for people to find you in the breaks.

Detailed schedule and program will be announced when available.

There will be four days of technical program and three days of cultural program:

All short presentations will be together with the poster session on Monday afternoon.

If you don't have time for the cultural program, you can arrive on Monday and leave on Friday; you will miss some cultural activities.

See detailed program.

 NEW  The oral session papers and some of the poster session papers are in LNCS proceedings vol. 6608 and 6609 published online. Abstracts of all LNCS papers can be downloaded in one file. In this file, the page numbers at the right side in the ToC correspond to the printed book, and the numbers at the left side correspond to the page numbers in this file only. If you print this file, make sure to enable the option "Document and markup" in the Reader's Print dialogue.

Please see the detailed program announced. Please check the time of your talk and let us know in case of any problem. It can be updated; please check it later.

Please reserve your places for the tours and give us other necessary information by filling in the registration form.

 SEE TRANSPORTATION GUIDE AT THE LOCAL INFORMATION PAGE.

The official conference hotel is Villa Fontaine Kudanshita. It is convenient for the participants to stay in the same hotel, to facilitate informal interaction. Usually our participants form ad-hoc informal companies in the hotel reception to go to some restaurant, local sightseeing, etc. All cultural activities will start from the official hotel. Registration will most probably take place in the lobby of this hotel and not at the conference venue. In addition, this hotel provided discounted rates for CICLing participants. We suggest you not to book a different hotel.

IMPORTANT! Our official hotel, Villa Fontaine Kudanshita, requires you to book your room personally using  this hotel registration form . Please print it, fill in all fields, and either fax it to them or scan and send them by email; see the fax number and the address in the form. (No, this could not be done via web form -- the hotel requires you to do it personally.) The deadline has passed, but still try submitting the form and contacting the hotel: they will do their best, though they cannot guarantee availability of rooms. In case of problems please contact us -- we will try to arrange for a workaround.

Please note that the rooms have one double bed. If you want to share a room with a twin bed, please let us know and we will try to arrange for a discounted rate at another hotel near the official one. In this case please do not send the form to Villa Fontaine Kudanshita; please contact us instead.

The rate is approximately 110 USD -- the best we could get (Japan is very expensive). The rate includes breakfast. There is wired internet in each room (they provide the cable; no wireless). There is a small bath tub in the room, a night pajama set (two if double occupancy), slippers, a TV set, air conditioner, and a kind of ironing device (pants press for ironing trousers and skirts). Rooms are no-smoking. It seems you cannot invite to your room guests not registered in the hotel. The hotel is a large modern building. The rooms are very small but clean and nice -- I've seen the rooms, here are my photos (click to enlarge).

Bed		Desk & TV		Desk & TV
Pajama		Cabled Internet		Window view (not all rooms)

Small bath tub		Pants press

The hotel is located in the very center of the city, almost next to the Emperor's palace—you will be in good company!  The downside is, it is not within walking distance from the conference venue. You will have to use the subway (3 stations to the conference site). We plan to provide you with a value-store card for the trip. The total time to the conference venue is about 20+ minutes: about 5 min walk to Kudanshita station (T-07), an average 2 minute wait for the next train, 6 min in the train to Waseda station (T-04), and 10 min walk. This is the closest hotel with reasonable rate and reliable internet connection, and a very centric location for sightseeing.

Instructions on getting to the hotel and to the conf venue from the hotel will be published here.

See additional information on the venue info page.