my photo

Amir H. Jadidinejad

Machine Learning Engineer

Learning from large scale textual corpora and its applications in Information Retrieval and Natural Language Processing are my favorite research interests. It's an emerging hot topic in the intersection of IR, NLP and ML. In particular, I am interested in deep learning models that go beyond the typical search engines to teach them how to understand text documents and user's information needs in order to search smarter. I am also interested in understanding how current models scale to very large collections, such as WWW. Learning conceptual representation of textual documents is a follow up research work. Currently, I specialize in the area of sequence to sequence models. I am interested in the following topics:

  • Deep Learning:
    • sequence to sequence models
    • encoder-decoder models
    • neural machine translation
    • semantic embedding
    • multi-task transfer learning
  • Information Retrieval:
    • web document ranking
    • query reformulation
    • understanding user's query intent
    • clustering/classification
  • Natural Language Engineering:
    • dialog systems
    • machine translation/transliteration
    • semantic relatedness
    • topic modeling
Contact me

Experience

Teaching

I teach the following courses at Qazvin Islamic Azad University from 2008 to now:

  • 2015-Now, Advanced topics in Information Retrieval (graduate).
  • 2015-Now, Advanced topics in Natural Language Processing (graduate).
  • 2011-Now, Introduction to Information Retrieval (undergraduate).
  • 2010-Now, Data Structure (undergraduate).
  • 2009-2015, Advanced Object-Oriented programming (undergraduate).
  • 2008-2009, Introduction to computer and programming (undergraduate).

Education

Islamic Azad University, Science and Research Branch

Ph.D. degree in software engineering, Tehran, Iran. • 2009 — 2015

Specialization in conceptual language processing, Dissertation topic: Knowledge-based feature generation for textual information based on constructing and analyzing semantic complex networks.

Qazvin Islamic Azad University

M.Sc. degree in software engineering, Qazvin, Iran. • 2007 — 2009

Specialization in Topic Modeling from Wikipedia, Dissertation topic: Mining Structured Information from Unstructured Texts using Wikipedia-based Semantic Analysis, GPA: 19.11/20 (top student).

Bu-Ali Sina University

B.Sc. degree in software engineering, Hamadan, Iran. • 2002 — 2006

Specialization in Document Clustering, Dissertation topic: Document Clustering Based on Fuzzy Approaches in Two-Stage Information Retrieval.

Projects

Deep Learning

Learning conceptual representation of textual documents • 2015 — Now

Studied on different applictions of deep learning (specially sequence-to-sequence models) in Information Retrieval and Natural Language Processing.

  • Amir H. Jadidinejad, ``Neural Machine Transliteration'', ArXive, 2016.

Conceptual Feature Generation

PhD thesis • 2010 — 2015

Studied on knowledge extraction from large scale Collaboratively Constructed Semantic Resources (specially Wikipedia as a multilingual, huge, popular encyclopedia and Wiktionary as Wikipedia's lexical companion) to construct a huge scale, machine-readable ontology and leverage it to solve some preliminary problems in IR and NLP. I specialized in the problem of mapping a fragment of unstructured text into Wikipedia Concept Space. This approach has been applied in different IR and NLP tasks such as: semantic similarity, document clustering/classification and etc.

  • Amir H. Jadidinejad, Fariborz Mahmoudi and M.R. Meybodi, ``Conceptual feature generation for textual information using a conceptual network constructed from Wikipedia'', Expert Systems, 33 (1), 92-106, 2016.
  • Amir H. Jadidinejad, Fariborz Mahmoudi and M.R. Meybodi, ``Clique-based semantic kernel with application to semantic relatedness'', Natural Language Engineering, 21 (5), 725-742, 2015.
  • Amir H. Jadidinejad, Fariborz Mahmoudi, ``Unsupervised Short Answer Grading Using Spreading Activation over an Associative Network of Concepts/La notation sans surveillance des rĂ©ponses courtes en utilisant la diffusion d'activation dans un rĂ©seau associatif de concepts'', Canadian Journal of Information and Library Science, 38 (4), 287-303, 2014.
  • Amir H. Jadidinejad, ``Unsupervised Information Extraction using BabelNet and DBpedi'', 3rd workshop on Making Sense of Microposts, World Wide Web Conference, Brazil, 2013.

Machine Learning in multidisciplinary fields

In collaboration with S.M.R Tousi • 2007 — 2008

Advisor of a Ph.D dissertation in Power Engineering in order to create an intelligent multi-agent framework to enhance elimination of voltage disturbances in power system contingencies.

  • MR Tousi, SH Hosseinian, AH Jadidinejad, MB Menhaj, ``Multi-agent based voltage control of STATCOMs to enhance elimination of voltage disturbances in power system contingencies'', In Proceeding of 43rd International Universities Power Engineering Conference, 2008.
  • M.R. Tousi, A.H. Jadidinejad, S.H. Hosseinian, M.B. Menhaj, ``Application of SARSA learning algorithm for reactive power control in power system'', In Proceeding of PECon 2008, 2nd IEEE International Conference on Power and Energy, Johor Bahru, Malaysia, December 2008.

National Referenced Health Semantic Model

Tamin Company • 2009 — 2010

Developed the first phase of standardization of electronic health record within a professional team composed of expert peoples in the field of computer engineering and medicine.

Persian Information Retrieval

Msc thesis • 2008 — 2010

Developed different tools and evaluated state-of-the-art structural query construction and retrieval models.

  • Amir H. Jadidinejad, Fariborz Mahmoudi and Jon Dehdari, ``Evaluation of Perstem: A Simple and Efficient Stemming Algorithm for Persian'', In Proceeding of the Multilingual Information Access Evaluation I. Text Retrieval Experiments, Lecture Notes in Computer Science, Volume 6241/2010, pp. 98-101, 2010.
  • Amir H. Jadidinejad, Fariborz Mahmoudi, ``Cross-Language Information Retrieval Using Meta-language Index Construction and Structural Queries'', In Proceeding of the Multilingual Information Access Evaluation I. Text Retrieval Experiments, Lecture Notes in Computer Science, Volume 6241/2010, pp. 70-77, 2010.
  • Amir H. Jadidinejad, Mitra Mohtarami, Hadi Amiri, ``Investigation on Application of Local Cluster Analysis and Part of Speech Tagging on Persian Text'', In Proceeding of the Cross Language Evaluation Forum (CLEF 2008 Working Notes), Aarhus, Denmark, 2008.
  • Amir H. Jadidinejad and Hadi Amiri, ``Local Cluster Analysis as a Basis for High-Precision Information Retrieval'', In Proceeding of INFOS2008 International Conference on Informatics and Systems, Cairo, Egypt, March 2008.

Skills

Programming

Python, Java, C/C++, R, Perl

Machine Learning Tools

Scikit-Learn, Weka, GenSim, GloVe, Mallet

Deep Learning Tools

TensorFlow, Theano

Information Retrieval Tools

Lemur, Indri, Terrier, Apache (Solar, Nutch, Lucene)

Natural Language Processing Tools

S-Space, Apache (Tika, Jena), KEA, NLTK

Development Tools

GIT, SVN, IPython Notebook, IntelliJ IDEA, Eclipse

Data bases

Neo4j, MySQL, MongoDB

Knowledge bases

BabelNet, DBpedia, Freebase