Hello, I am Jonathan. I am final year PhD student at the University of Edinburgh under the supervision of Mirella Lapata and Rico Sennrich. My main interest is text generation; in particular paraphrasing and sentence compression. I am funded by EPSRC's CDT in Data Science program. Before coming to Edinburgh I was at the University of Amsterdam where I did a masters in logic. Previously, I studied Computer Science at the Unversity of Birmingham.
|2018|| Sentence Compression for Arbitrary Languages via Multilingual Pivoting.
Jonathan Mallinson; Rico Sennrich; Mirella Lapata (2018). Accepted into the Conference on Empirical Methods in Natural Language Processing; Brussels, Belgium.
|2017|| Paraphrasing Revisited with Neural Machine Translation.
Jonathan Mallinson; Rico Sennrich; Mirella Lapata (2017). In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics, Valencia, Spain.
|2017||Learning Paraphrastic Sentence Embeddings from Back-Translated Bitext
John Wieting; Jonathan Mallinson; Kevin Gimpel (2017). Accepted into the Conference on Empirical Methods in Natural Language Processing; Copenhagen, Denmark.
|2017||Learning to Paraphrase for Question Answering
Li Dong; Jonathan Mallinson; Siva Reddy; Mirella Lapata (2017). Accepted into the Conference on Empirical Methods in Natural Language Processing; Copenhagen, Denmark.
|2016||PARANET: Bilingual Encoder-Decoder Paraphrasing
MSc(R) in Data Science.
I explored in detail how neural machine translation can be used to paraphrase. This dissertation formed the basis of my paper "Paraphrasing Revisited with Neural Machine Translation". Supervisor: Mirella Lapata
|2015||Modelling Syntactic and Semantic Tasks with Linguistically Enriched Recursive Neural Networks
MSc in Logic.
Recursive Neural Networks, enhanced with linguistically motivated features were successfully used as a syntactic parser and paraphrase detector. Supervisor: Willem (Jelle) Zuidema
|2011||Simulated Stock Market using Fuzzy Agents
BSc in Computer Science.
Multi-agent modeling of the stock market, agents used fuzzy logic for decision making. Evolutionary learning was applied to minimise the objective function. Supervisor: Antoni Diller
|2015||Sentiment analysis using flat trees|
|2015||Unsupervised semantic role labelling with constraints
Expectation maximization (EM) was used to train HMM parametrised with features, under constraints imposed by VerbNet, to learn semantic roles in an unsupervised approach. Supervisor: Ivan Titov
|2014||Predicting the meaning in use of occurrences of Fall
Working in a team of researchers I helped to provide a data-based semantic/pragmatic analysis of "FALL". Supervisor: Henk Zeevat
|2012||Lexical simplification - Data Augmentation
Augmentation of the Semeval 2012 lexical simplification dataset was found to improve upon the baseline system. Supervisor: Raquel Fernández
|2018|| Intern: Allen Institute for Artificial Intelligence (AI2)
As an intern, I was a member of the AllenNLP team, where I was primarily supervised by Mohit Iyyer
Received an AI3 award
|2017 onwards|| Teaching assistant, Demonstrator, Marker: Machine translation at the University of Edinburgh
Nominated for a Teaching Support Award
I'm currently enjoying my final year of my PhD.
I completed a one year research masters in Data Science. My studies focused on a mix of Machine learning and NLP.
I took a Masters in Logic, taking the "logic & language" track. Which included learning a mix of semantics, NLP and linguistics.
I spent three years at the University of Birmingham studying Computer Science.