Dmitry Davidov
Ph.D Student
Hebrew University of Jerusalem
Interdisciplinary Center for Neural Computation (ICNC)
Jerusalem [Israel]
dmitry(~@~)alice.nc.huji.ac.il
Phone: (+972) 544-522-793
Supervisor:
Dr. Ari Rappoport

 

Navigation

 

Introduction

Research interests

Publications

Introduction

 

I am a Ph.D. student at Hebrew University of Jerusalem, working on semantic acquisition                                     

I have received my M.A. at Technion  (2003), focusing on heuristics and AI.

I am also working as a programmer at the Motor Cortex research laboratory.

 

Research Interests

 

My main research interest is developing a framework for fully unsupervised and maximally language-independent acquisition of semantic concepts and relationships.

Additional interests include ontology construction, first language acquisition, and heuristic search algorithms.

 

Publications

 

Unsupervised Semantic acquisition:

Dmitry Davidov, Roi Reichart, Ari Rappoport. Superior and Efficient Fully Unsupervised Pattern-based Concept Acquisition Using an Unsupervised Parser. Computational Natural Language Learning (CoNLL) 2009. (pdf)

-This work deals with incorporation of unsupervised parser into fully unsupervised concept acquisition framework

 

Elad Dinur, Dmitry Davidov, Ari Rappoport. Unsupervised Concept Discovery In Hebrew Using Simple Unsupervised Word Prefix Segmentation for Hebrew and Arabic. EACL 2009 Workshop on Computational Approaches to Semitic Languages (pdf)

-This work describes incorporation of unsupervised word prefix segmentation into fully unsupervised concept acquisition framework

 

Dmitry Davidov, Ari Rappoport. Classification of Semantic Relationships between Nominals Using Pattern Clusters. Accepted as a full oral presentation, ACL 2008. (pdf)

-This work describes application of unsupervisingly acquired pattern clusters to nominal classification

 

Dmitry Davidov, Ari Rappoport. Unsupervised Discovery of Generic Relationships Using Pattern Clusters and its Evaluation by Automatically Generated SAT Analogy Questions. Accepted as a full oral presentation, ACL 2008. (pdf)

-This work presents a way to define unspecified relationships as pattern clusters and to obtain such clusters automatically from corpus without relying on any kind of language-specific information or training data.

 

Dmitry Davidov, Ari Rappoport, Moshe Koppel . Fully Unsupervised Discovery of Concept-Specific Relationships by Web Mining. Proceedings, ACL 2007, June 2007, Prague. (pdf)

- This work describes web-based discovery and extraction of unspecified concept-specific relationships from web starting with a few seed words as concept definition.

 

Dmitry Davidov, Ari Rappoport. Efficient Unsupervised Discovery of Word Categories Using Symmetric Patterns and High Frequency Words. COLING-ACL 06.(pdf)

- This work describes completely unsupervised pattern-based acquisition of lexical categories from text corpora. The method based solely on symmetry property of language and does not require any word/pattern seeds.

 

 

Multilinguality:

Dmitry Davidov, Ari Rappoport. Enhancement of Lexical Concepts Using Cross-lingual Web Mining. EMNLP 2009. (pdf)

-This work presents a way to enhance lexical concepts by discovery and utilization of the information available for the given concept in many languages

 

Dmitry Davidov, Ari Rappoport.. Translation and Extension of Concepts Across Languages. EACL 2009. (pdf)

-This work describes fully automated and language-independent translation and extension of concepts evaluated on dozens of diverse languages

 

 

Discovery of specific relationship types:

Dmitry Davidov, Ari Rappoport. Geo-mining: Discovery of Road and Transport Networks Using Directional Patterns. EMNLP 2009. (pdf)

-This paper presents an an algorithmic framework which allows an automated acquisition of map-like information from the web, based on surface patterns like “from X to Y”

 

 

Linguistic data sets generation:

Dmitry Davidov, Evgeniy Gabrilovich, Shaul Markovitch: Parameterized generation of labeled datasets for text categorization based on a hierarchical directory. SIGIR 2004: 250-257(pdf)

- This work describes automated generation of data sets for text classification based on open directory database.

 

 

Heuristics & algorithms:

Dmitry Davidov and Shaul Markovitch. Multiple-goal Heuristic Search. Journal of Artificial Intelligence Research, 26:417-451, 2006[pdf]

- This work includes extensive research on multiple-goal search algorithms, their performance and applications

 

Dmitry Davidov, Shaul Markovitch: Multiple-Goal Search Algorithms and their Application to Web Crawling. AAAI/IAAI 2002: 713-718.[pdf ]

- This work develops and evaluates heuristics and algorithms for multiple-goal graph search when many goal nodes are to be collected simultaneously.

 

My M.A. thesis on Multiple Goal Search algorithms [pdf,English] and its Hebrew summary[doc ]