The OPT submission to the Shared Task of the 2016 Conference on Natural Language Learning (CoNLL) implements a 'classic' pipeline architecture, combining binary classification of (candidate) explicit connectives, heuristic rules for non-explicit discourse relations, ranking and 'editing' of syntactic constituents for argument identification, and an ensemble of classifiers to assign discourse senses. With an end-toend performance of 27.77 F 1 on the English 'blind' test data, our system advances the previous state of the art (Wang & Lan, 2015) by close to four F 1 points, with particularly good results for the argument identification sub-tasks.

This paper presents an alternative approach to polarity and intensity classification of sentiments in microblogs. In contrast to previous works, which either relied on carefully designed hand-crafted feature sets or automatically derived neural embeddings for words, our method harnesses character embeddings as its main input units. We obtain task-specific vector representations of characters by training a deep multi-layer convolutional neural network on the labeled dataset provided to the participants of the SemEval-2016 Shared Task 4 (Sentiment Analysis in Twitter; Nakov et al., 2016b) and subsequently evaluate our classifiers on subtasks B (two-way polarity classification) and C (joint five-way prediction of polarity and intensity) of this competition. Our first system, which uses three manifold convolution sets followed by four non-linear layers, ranks 16 in the former track; while our second network, which consists of a single convolutional filter set followed by a highway layer and three non-linearities with linear mappings in-between, attains the 10-th place on subtask C.

This paper addresses the problem of segmenting German texts into minimal discourse units, as they are needed, for example, in RST-based discourse parsing. We discuss relevant variants of the problem, introduce the design of our annotation guidelines, and provide the results of an extensive interannotator agreement study of the corpus. Afterwards, we report on our experiments with three automatic classifiers that rely on the output of state-of-the-art parsers and use different amounts and kinds of syntactic knowledge: constituent parsing versus dependency parsing; tree-structure classification versus sequence labeling. Finally, we compare our approaches with the recent discourse segmentation methods proposed for English.