Difference between revisions of "Research"

(Bayesian Modeling)
 
(33 intermediate revisions by 6 users not shown)
Line 1: Line 1:
 +
In the CLIP lab, we approach research on computational linguistics and information processing from a variety of angles. Some of our ongoing projects focus on the following challenges:
  
 +
* Computational psycholinguistics
 +
* Computational social science
 +
* Cross-language information retrieval
 +
* Data science for finance / social good
 +
* Deep learning
 +
* E-discovery
 +
* Pattern discover in graphs / ranking and recommendation
 +
* Human-in-the-loop machine learning
 +
* Machine translation
 +
* Mental health
 +
* Privacy-aware information retrieval
 +
* Speech retrieval
 +
* Urban computing / smart environments
  
=Bayesian Modeling=
+
CLIP research has been supported by the following organizations: NSF, DARPA, ARL, IARPA, OFR (Treasury), NIST, IMLS, Google, Yahoo and the World Bank.
 
 
{|
 
|-
 
|  | <b>Faculty</b>
 
|  [http://www.umiacs.umd.edu/~jbg Jordan Boyd-Graber], [http://terpconnect.umd.edu/~naomi Naomi Feldman], [http://umiacs.umd.edu/~hal Hal Daum&eacute; III], [http://www.umiacs.umd.edu/~resnik/ Philip Resnik]
 
|-
 
|  | <b>Postdocs </b>
 
| [http://www.umiacs.umd.edu/~tsmoon/ Taesun Moon]
 
|-
 
|  <b>Graduate Students </b>
 
|  [http://www.cs.umd.edu/~vietan/ Viet-An Nguyen] [http://umiacs.umd.edu/~ynhu/ Yuening Hu], Ke Zhai
 
|}
 
 
 
=== What We Do ===
 
 
 
Bayesian modeling is a rigorous mathematical formalism that allows us to build systems that reflect our uncertainty about the world.  Applied to language, they allow us to build models that reflect the "latent" aspects of communication such as topic, part of speech, syntax, or sentiment.  Using posterior inference, we can use the models to discover the latent features that best explain observed language.
 
 
 
In the CLIP lab, we are interested in
 
* building tools that make it easier for people to work with Bayesian models
 
* scaling inference for Bayesian models up to Web scale
 
* understanding how humans interpret and understand the latent variables in Bayesian models
 
 
 
=== Publications ===
 
 
 
* [http://www.umiacs.umd.edu/~hal/HBC/ Hierarchical Bayes Compiler]
 
* [http://www.umiacs.umd.edu/~jbg/docs/nips2009-rtl.pdf Reading Tea Leaves: How Humans Interpret Topic Models]
 
* [http://drum.lib.umd.edu/handle/1903/10058 Gibbs Sampling for the Uninitiated]
 
 
 
==Machine Translation==
 
 
 
{| border="0" cellpadding="5" cellspacing="0" align="center"
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Faculty</b>
 
|
 
{| border="0"
 
| [http://www.umiacs.umd.edu/~bonnie Bonnie Dorr] || interlingual and hybrid machine translation, MT evaluation
 
|-
 
| [http://umiacs.umd.edu/~mharper Mary Harper] || multilingual parsing, language modeling
 
|-
 
| [http://www.umiacs.umd.edu/~resnik/ Philip Resnik] || linguistically informed translation modeling, crowdsourcing and translation
 
|-
 
| [http://www.umiacs.umd.edu/~hal/ Hal Daum&eacute; III] || domain adaptation for translation; translation with linguistic universals
 
|-
 
|}
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Postdocs </b>
 
|  | [http://www.umiacs.umd.edu/~hollingk/ Kristy Hollingshead]
 
|-
 
| style="border-bottom: 3px solid grey; border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Graduate Students </b>
 
| style="border-bottom: 3px solid grey;" | [http://www.cs.umd.edu/~vlad/ Vladimir Eidelman]
 
|-
 
| colspan="3" align="left" |
 
 
 
The CLIP Laboratory's current work in <b>machine translation</b> continues the lab's long tradition of research in this area.  Like most of the field, we work within the framework of statistical MT, but with an emphasis on taking appropriate advantage of knowledge driven or linguistically informed model structures, features, and priors.  Some current areas of research include syntactically informed language models, linguistically informed translation model features, the use of unsupervised methods in translation modeling, exploitation of large scale "cloud computing" methods, and human-machine collaborative translation via crowdsourcing.
 
 
 
<b>Some Representative Publications:</b>
 
* Chris Dyer and Philip Resnik. [http://www.aclweb.org/anthology/N/N10/N10-1128.pdf Context-free reordering, finite-state translation]. In Proceedings of NAACL-HLT 2010, Los Angeles, CA, USA, 2010.
 
* Hendra Setiawan, Chris Dyer, and Philip Resnik. [http://www.cs.cmu.edu/~cdyer/ Discriminative Word Alignment with a Function Word Reordering Model].  Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), Cambridge, MA, USA, 2010.
 
* Matt Snover, Nitin Madnani, Bonnie Dorr, and Richard Schwartz, [ftp://ftp.umiacs.umd.edu/pub/bonnie/10590_2009_9062_Author.pdf TER-Plus: Paraphrases, Semantic, and Alignment Enhancements to Translation Edit Rate], Machine Translation, 23:2-3, Springer Netherlands, pp. 117-127, 2009.
 
 
 
<b>Some Project Pages</b>
 
* [http://www.cs.umd.edu/hcil/monotrans/ Monotrans: Translation Crowdsourcing without Bilinguals]
 
 
 
==Paraphrase==
 
 
 
{| border="0" cellpadding="5" cellspacing="0" align="center"
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Faculty</b>
 
|
 
{| border="0"
 
| [http://www.umiacs.umd.edu/~bonnie Bonnie Dorr] || paraphrasing, summarization, language understanding
 
|-
 
| [http://www.umiacs.umd.edu/~resnik/ Philip Resnik] || linguistically informed NLP, paraphrasing
 
 
 
|-
 
|}
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Postdocs </b>
 
|
 
|-
 
| style="border-bottom: 3px solid grey; border-right: 1px solid grey; background:#ffefef" valign="top" | <b> Students </b>
 
| style="border-bottom: 3px solid grey;"  | Olivia Buzek, [http://www.ling.umd.edu/~yakov/ Yakov Kronrod]
 
| colspan="3" align="left" |
 
|-
 
 
 
 
 
<b>Paraphrase</b>, the ability to express the same meaning in multiple ways, is an active area of research within the NLP community and here in the CLIP Laboratory.  Our work in paraphrase  includes the use of paraphrase in MT evaluation and parameter estimation, lattice and forest translation, and collaborative translation, as well as research on lexical and phrasal semantic similarity measures, meaning preservation in machine translation and summarization, and large-scale document similarity computation via cloud computing methods. 
 
 
 
|}
 
 
 
<b>Some Project Pages</b>
 
* [http://www.cs.umd.edu/hcil/monotrans/ Monotrans: Crowdsourcing Translation without Bilinguals]
 
 
 
<b>Some Representative Publications</b>
 
* Generating Phrasal & Sentential Paraphrases: A Survey of Data-Driven Methods. 2010. Computational Linguistics, 36(3). Nitin Madnani and Bonnie Dorr.
 
* Philip Resnik, Olivia Buzek, Chang Hu, Yakov Kronrod, Alex Quinn, Benjamin B. Bederson. [http://ling.umd.edu/~yakov/CrowdConf2010/final.pdf Improving Translation via Targeted Paraphrasing], 2010 Conference on Empirical Methods in Natural Language Processing, October 2010.
 
* Yuval Marton, Saif Mohammad, and Philip Resnik. [http://www.aclweb.org/anthology/D/D09/D09-1081.pdf Estimating Semantic Distance Using Soft Semantic Constraints in Knowledge-Source / Corpus Hybrid Models']. Conference on Empirical Methods in Natural Language Processing (EMNLP). Singapore, August 6-7, 2009.
 
* Nitin Madnani, Necip Fazil Ayan, Philip Resnik, Bonnie Dorr. [http://www.desilinguist.org/pdf/paraphrase-wmt07.pdf Using Paraphrases for Parameter Tuning in Statistical Machine Translation]. 2007. Proceedings of the Second ACL Workshop on Statistical Machine Translation (WMT-07).
 
 
 
==Text Summarization ==
 
 
 
{| border="0" cellpadding="5" cellspacing="0" align="center"
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Faculty</b>
 
|
 
{| border="0"
 
| [http://www.umiacs.umd.edu/~bonnie Bonnie Dorr] || evaluation
 
|-
 
| David Zajic || sentence compression, sentence selection
 
|-
 
| [http://www.umiacs.umd.edu/~hal Hal Daum&eacute; III] || summarization of technical documents; sentence compression
 
|-
 
|}
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Postdocs </b>
 
|
 
|-
 
| style="border-bottom: 3px solid grey; border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Graduate Students </b>
 
| style="border-bottom: 3px solid grey;" |
 
|-
 
| colspan="3" align="left" |
 
 
 
Text Summarization is the creation of a short document to serve as a surrogate for a longer document.  The CLIP Laboratory's approach to summarization enhances the extractive method of selecting source document sentences for inclusion in a summary by using sentence compression to enlarge the pool of available sentences, and by combining fluent text with topic terms.  Our sentence compression technology has encompassed both statistical and linguistic methodologies.  We have developed an extrinsic evaluation measure for summarization, Relevance Prediction, which is grounded in a real-world task using summarized documents. The CLIP Laboratory, in collaboration with BBN, has been a regular participant in NIST's summarization evaluations (Document Understanding Conferences and Text Analysis Conferences), and has contributed summarization components to DARPA Translingual Information Detection, Extraction and Summarization (TIDES), Surprise Language Exercise (SLE), and Global Autonomous Language Exploitation (GALE) programs, and to the iOpener project.
 
 
 
|}
 
 
 
==Parsing and Tagging==
 
 
 
{| border="0" cellpadding="5" cellspacing="0" align="center"
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Faculty</b>
 
|
 
{| border="0"
 
| [http://www.umiacs.umd.edu/~jbg/ Mary Harper] || latent variable parsing, speech
 
|-
 
|}
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Postdocs </b>
 
|
 
|-
 
| style="border-bottom: 3px solid grey; border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Graduate Students </b>
 
| style="border-bottom: 3px solid grey;" | [http://www.umiacs.umd.edu/~vlad/ Vladimir Eidelman]  [http://www.umiacs.umd.edu/~zhuang/ Zhongqiang Huang]
 
|-
 
| colspan="3" align="left" |
 
 
 
<b>Parsing</b> involves
 
 
 
<b>Representative Publications and Project Pages:</b>
 
* Huang and Harper, EMNLP 2009: [http://www.aclweb.org/anthology/D/D09/D09-1087.pdf Self-Training PCFG Grammars with Latent Annotations Across Languages]
 
* Huang, Eidelman and Harper NAACL 2009: [http://www.umiacs.umd.edu/~vlad/papers/tagger-la-st_naacl09.pdf Improving A Simple Bigram HMM Part-of-Speech Tagger by Latent Annotation and Self-Training ]
 
 
 
|}
 
 
 
==Computational Social Science==
 
 
 
 
 
{| border="0" cellpadding="5" cellspacing="0" align="center"
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Faculty</b>
 
|
 
{| border="0"
 
| [http://www.umiacs.umd.edu/~jbg/ Jordan Boyd-Graber] || scientific literature analysis, persuasion
 
|-
 
| [http://www.umiacs.umd.edu/~bonnie Bonnie Dorr] || sentiment analysis, scientific literature analysis
 
|-
 
| [http://www.umiacs.umd.edu/~jimmylin/ Jimmy Lin] || social media
 
|-
 
|  [http://www.umiacs.umd.edu/~oard/ Douglas W. Oard] || topical relation detection
 
|-
 
| [http://www.umiacs.umd.edu/~louiqa/ Louiqa Raschid] || diffusion, prediction, event detection, recommendation
 
|-
 
| [http://www.umiacs.umd.edu/~resnik/ Philip Resnik] || sentiment, persuasion
 
|-
 
| [http://www.umiacs.umd.edu/~weinberg Amy Weinberg] || sentiment, persuasion
 
|}
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Postdocs </b>
 
| [http://www.cs.umd.edu/~skok/ Stanley Kok]
 
|-
 
| style="border-bottom: 3px solid grey; border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Graduate Students </b>
 
| style="border-bottom: 3px solid grey;" | [http://www.umiacs.umd.edu/~asayeed/ Asad Sayed] [http://www.cs.umd.edu/~sayyadi Hassan Sayyadi] [http://www.cs.umd.edu/~wsc Shanchan Wu]
 
|-
 
| colspan="3" align="left" |
 
 
 
<b>Computational social science</b> involves the use of computational methods and models to leverage [http://www.sciencemag.org/cgi/content/summary/323/5915/721 "the capacity to collect and analyze data at a scale that may reveal patterns of individual and group behaviors"].  Research in the CLIP Laboratory is at the forefront of this emerging area, and includes sentiment analysis (computational modeling and prediction of opinions, perspective, and other private states), automatic analysis and visualization of the scientific literature, modeling the diffusion of technological innovations, modeling and prediction of social goals and actions such as persuasion, monitoring and prediction (tracking events, predicting new links or articles) and recommendation (personalized recommendations, learning to rank).
 
 
 
<b>Representative Publications and Project Pages:</b>
 
* Greene and Resnik, NAACL 2009: [http://umiacs.umd.edu/~resnik/pubs/greene_resnik_naacl2009.pdf More Than Words: Syntactic Packaging and Implicit Sentiment]
 
* Boyd-Graber and Resnik, EMNLP 2010: [http://www.umiacs.umd.edu/~jbg/docs/jbg-mlslda-2010.pdf Holistic Sentiment Analysis Across Languages: Multilingual Supervised Latent Dirichlet Allocation]
 
* Joshi, Y., Rand, W. and Raschid, L. [http://www.rhsmith.umd.edu/ccb/hccgrant.aspx Diffusion and Ranking in Digital Social Media]
 
 
 
|}
 
 
 
==Information Retrieval ==
 
 
 
{| border="0" cellpadding="5" cellspacing="0" align="center"
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Faculty</b>
 
|
 
{| border="0"
 
| [http://www.umiacs.umd.edu/~jimmylin/ Jimmy Lin] ||
 
|-
 
| [http://terpconnect.umd.edu/~oard/ Douglas W. Oard] ||
 
|-
 
|}
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Postdocs </b>
 
|
 
{| border="0"
 
| [http://www.umiacs.umd.edu/~ewagner/ Earl Wagner] ||
 
|-
 
[http://www.umiacs.umd.edu/~wew/ William Webber] ||
 
|-
 
|}
 
|-
 
| style="border-bottom: 3px solid grey; border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Graduate Students </b>
 
| style="border-bottom: 3px solid grey;" | [http://ar-ar.facebook.com/bagdouri Mossaab Bagdouri],  [https://plus.google.com/101329784257647049204/posts Sergey Golitsynskiy], [http://iitgalumni.org/govind88 Govind Kothari], [http://advancement.umd.edu/celebration/showStudent.php?main_id=3193 Levon Mkrtchyan], [http://www.cs.umd.edu/~fture/ Ferhan Ture], [http://www.cs.umd.edu/~lidan Lidan Wang], [http://terpconnect.umd.edu/~tanx/ Tan Xu]
 
|- [http://www.umiacs.umd.edu/~ewagner/ Earl Wagner] ||
 
| colspan="3" align="left" |
 
 
 
The goal of information retrieval is to help people find what they are looking for.  Information retrieval research in the CLIP lab focuses principally on retrieval based on the language contained in text, in speech, and in document images.  We work across a broad range of content types, from tweets to tomes, from talking to texting, and from Cebuano to Chinese.  Three perspectives inform our work:
 
* we integrate a broad range of computational linguistics techniques,
 
* we focus on scalable techniques that can accommodate very large collections
 
* we sometimes draw the boundaries of our “systems” very broadly to include both the automated tools that we create and the process by which users can best employ those tools.
 
 
 
One example that illustrates these perspectives is our work with “cross-language information retrieval,” in which close coupling of machine translation and information retrieval techniques make it possible for people to find and use information written in languages that they can neither read nor write.  Another example is our work on the design and evaluation of “question answering” systems that can automatically find and present answers to complex questions, which serves as a bridge between our work on information retrieval and summarization.
 
 
 
<b>Representative Publications and Project Pages:</b>
 
* Douglas W. Oard, "[http://terpconnect.umd.edu/~oard/pdf/elis09.pdf Multilingual Information Access]," in Encyclopedia of Library and Information Sciences, 3rd Ed., 2009.
 
* Project: [http://ediscovery.umiacs.umd.edu/ Development and Evaluation of Search Technology for Discovery of Evidence in Civil Litigation]
 
 
 
 
 
|}
 
 
 
==Disambiguation==
 
 
 
{| border="0" cellpadding="5" cellspacing="0" align="center"
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Faculty</b>
 
|
 
{| border="0"
 
| [http://www.umiacs.umd.edu/~jbg Jordan Boyd-Graber] ||
 
|-
 
| [http://www.umiacs.umd.edu/~jklavans Judith Klavans] ||
 
|-
 
| [http://www.umiacs.umd.edu/~resnik/ Philip Resnik] ||
 
|-
 
|}
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Postdocs </b>
 
|
 
|-
 
| style="border-bottom: 3px solid grey; border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Graduate Students </b>
 
| style="border-bottom: 3px solid grey;" | Raul David Guerra
 
|-
 
|  colspan="3" align="left" |
 
 
 
<b>Disambiguation </b> is the process of determining the meaning or senses of a word in its context;  disambiguation remains one of the most challenging NLP problems since discovering word senses involves syntactic, semantic and pragmatic contextual inferencing, along with a rich knowledge base to base selection upon. For example, the word "wing" in the theater differs from airplanes, yet another sense for furniture ("wing chair") applies to some usages.  Often disambiguation can be based on windows of two and three words, but usually involves larger computation.  Techniques for disambiguation range from the use of large scale thesaural resources (such as WordNet) to purely statistical methods.
 
 
 
<b>Representative Publications and Project Pages:</b>
 
* [http://www.umiacs.umd.edu/research/t3/index.shtml T3: Text, Tags, Trust]
 
* [http://www.umiacs.umd.edu/~jbg/docs/jbg-EMNLP07.pdf A Topic Model for Word Sense Disambiguation]
 
* Philip Resnik and David Yarowsky, "[http://www.cs.jhu.edu/~yarowsky/pubs/nle00.ps Distinguishing Systems and Distinguishing Senses: New Evaluation Methods for Word Sense Disambiguation]", Natural Language Engineering 5(2), pp. 113-133.
 
 
 
|}
 
 
 
==Annotation and Sense-making==
 
 
 
{| border="0" cellpadding="5" cellspacing="0" align="center"
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Faculty</b>
 
|
 
{| border="0"
 
 
 
| [http://www.umiacs.umd.edu/~jklavans Judith Klavans] ||
 
|-
 
| [http://www.umiacs.umd.edu/~louiqa Louiqa Raschid] ||
 
|-
 
 
 
 
 
|-
 
|}
 
|-
 
| style="border-right: 1px solid grey; background:#ffefef" valign="top"  | <b>Postdocs </b>
 
|
 
|-
 
| style="border-bottom: 3px solid grey; border-right: 1px solid grey; background:#ffefef" valign="top" | <b>Graduate Students </b>
 
| style="border-bottom: 3px solid grey;" | [http://www.cs.umd.edu/~sayyadi Hassan Sayyadi] [http://www.cs.umd.edu/~wsc Shanchan Wu]
 
|-
 
| colspan="3" align="left" |
 
 
 
Annotation and tagging are ways to enhance knowledge in structured or semi-structured resources. Annotation typically references terms from a controlled vocabulary or ontology and is popular in bibliographic, scientific or museum collections. Tagging is more common in social media to tag images and documents and of course the now ubiquitous hashtags tweets. Sense-making or discovery is the process of extracting knowledge from these annotated or tagged resources and could range from simple counting to data/text mining to graph pattern recognition.
 
 
 
In the CLIP lab, we are interested in the following tasks:
 
* Tagging and sense-making
 
* Pattern discovery in annotated graph datasets from the biomedical domain.
 
* Data mining with Linked Data.
 
 
 
<b>Representative Publications and Project Pages:</b>
 
* [https://wiki.umiacs.umd.edu/clip/pattaran/index.php/Main_Page NSF DBI 1147144 Methodology for Pattern Creation, Imprint Validation, and Discovery from the Annotated Biological Web (PattArAn)]
 
* [http://www.umiacs.umd.edu/~louiqa/2012/RSEAGER2009.html Pattern Discovery, Validation, and Hypothesis Development from the Annotated Biological Web ]
 
 
 
|}
 
 
 
== Recent Accomplishments in the last 12 months ==
 
 
 
Jordan Boyd-Graber
 
 
 
Hal Daume III
 
 
 
David Doermann
 
 
 
Bonnie Dorr
 
 
 
Jimmy Lin
 
 
 
Doug Oard
 
 
 
Louiqa Raschid
 
PattArAn: NSF grant and collaboration with plant biologists.
 
SM3: NSF grant and multiple papers in collaboration with Shanchan Wu and Hassan Sayyadi and Bill Rand.
 
PAnG: Tool for graph data mining of annotated graph datasets; collaboration with Samir Khuller and multiple papers.
 
Next Generation Financial Cyberinfrastructure: Workshops in July 2010 and July 2012 sponsored by the NSF and CRA/CCC.
 
 
 
Philip Resnik
 

Latest revision as of 01:51, 8 September 2017

In the CLIP lab, we approach research on computational linguistics and information processing from a variety of angles. Some of our ongoing projects focus on the following challenges:

  • Computational psycholinguistics
  • Computational social science
  • Cross-language information retrieval
  • Data science for finance / social good
  • Deep learning
  • E-discovery
  • Pattern discover in graphs / ranking and recommendation
  • Human-in-the-loop machine learning
  • Machine translation
  • Mental health
  • Privacy-aware information retrieval
  • Speech retrieval
  • Urban computing / smart environments

CLIP research has been supported by the following organizations: NSF, DARPA, ARL, IARPA, OFR (Treasury), NIST, IMLS, Google, Yahoo and the World Bank.