AI-generated poetry for World Poetry Day 2019

What happens when artificial intelligence is challenged to write poetry? Recently, a team of creatives, developers and robotics specialists looked to the work of Sylvia Plath to find out.Bristol-based business Tiny Giant teamed up with Sharpshooter to train a recurrent neural network — a type of machine learning algorithm — on the poems of Sylvia Plath. They then joined forces with robotics experts, The Handwriting Company, to bring them to life.The machine-learning model generated hundreds of lines of Plath-inspired verse, which the team’s writers crafted into five unique poems. These were taken to The Handwriting Company…

Read More

Speech and Language Processing

(3rd ed. draft) Dan Jurafsky and James H. Martin Draft chapters in progress, Sep 23, 2018 This is the release for the start of fall term 2018. The slides are in the process of being updated now, we are putting them up as we write them.  Significantly rewritten version of 5, 6, 7, 8, 17, 18, 19, 23, 24, 25, and a draft of 9! New pedagogical sequences on neural networks and their training, starting with logistic regression and continuing with embeddings, feed-forward nets, and RNNs. Plus new or improved coverage of…

Read More

A Structured Review of the Validity of BLEU – Ehud Reiter,

Posted Online September 21, 2018 https://doi.org/10.1162/coli_a_00322 © 2018 Association for Computational Linguistics Published under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license Abstract The BLEU metric has been widely used in NLP for over 15 years to evaluate NLP systems, especially in machine translation and natural language generation. I present a structured review of the evidence on whether BLEU is a valid evaluation technique—in other words, whether BLEU scores correlate with real-world utility and user-satisfaction of NLP systems; this review covers 284 correlations reported in 34 papers. Overall,…

Read More

Computer-assisted reviewing (CAR)

Computer-assisted reviewing (CAR) tools are pieces of software based on text-comparison and analysis algorithms. These tools focus on the differences between two documents, taking into account each document’s typeface through an intelligent analysis. Detecting differences[edit] The intelligent analysis used by CAR tools detect the differences do not have the same value depending on their type and/or the document field/subject. For example, a difference on a number is not the same if this number is a date, a price, a page number, a figure number, a part of an address, a…

Read More