This paper describes EmbLexChange, a system introduced by the “Life-Language” team for SemEval-2020 Task 1, on unsupervised detection of lexical-semantic changes. EmbLexChange is defined as the divergence between the embedding based profiles of word w (calculated with respect to a set of reference words) in the source and the target domains (source and target domains can be simply two time frames t_1 and t_2). The underlying assumption is that the lexical-semantic change of word w would affect its co-occurring words and subsequently alters the neighborhoods in the embedding spaces. We show that using a resampling framework for the selection of reference words (with conserved senses), we can more reliably detect lexical-semantic changes in English, German, Swedish, and Latin. EmbLexChange achieved second place in the binary detection of semantic changes in the SemEval-2020.
In this paper, we introduce UniSent universal sentiment lexica for 1000+ languages. Sentiment lexica are vital for sentiment analysis in absence of document-level annotations, a very common scenario for low-resource languages. To the best of our knowledge, UniSent is the largest sentiment resource to date in terms of the number of covered languages, including many low resource ones. In this work, we use a massively parallel Bible corpus to project sentiment information from English to other languages for sentiment analysis on Twitter data. We introduce a method called DomDrift to mitigate the huge domain mismatch between Bible and Twitter by a confidence weighting scheme that uses domain-specific embeddings to compare the nearest neighbors for a candidate sentiment word in the source (Bible) and target (Twitter) domain. We evaluate the quality of UniSent in a subset of languages for which manually created ground truth was available, Macedonian, Czech, German, Spanish, and French. We show that the quality of UniSent is comparable to manually created sentiment resources when it is used as the sentiment seed for the task of word sentiment prediction on top of embedding representations. In addition, we show that emoticon sentiments could be reliably predicted in the Twitter domain using only UniSent and monolingual embeddings in German, Spanish, French, and Italian. With the publication of this paper, we release the UniSent sentiment lexica at http://language-lab.info/unisent.
For compounding languages, a great part of the topical semantics is transported via nominal compounds. Various applications of natural language processing can profit from explicit access to these compounds, provided by a lexicon. The best way to acquire such a resource is to harvest corpora that represent the domain in question. For Chinese, a significant difficulty lies in the fact that the text comes as a string of characters, only segmented by sentence boundaries. Extraction algorithms that solely rely on context variety do not perform precisely enough. We propose a pipeline of filters that starts from a candidate set established by accessor variety and then employs several methods to improve precision. For the experiments the Xinhua part of the Chinese Gigaword Corpus was used. We extracted a random sample of 200 story texts with 119,509 Hanzi characters. All compound words of this evaluation corpus were tagged, segmented into their morphemes, and augmented with the POS-information of their segments. A cascade of filters applied to a preliminary set of compound candidates led to a very high precision of over 90%, measured for the types. The result also holds for a small corpus where a solely contextual method introduces too much noise, even for the longer compounds. An introduction of MI into the basic candidacy algorithm led to a much higher recall with still reasonable precision for subsequent manual processing. Especially for the four-character compounds, that in our sample represent over 40% of the target data, the method has sufficient efficacy to support the rapid construction of compound dictionaries from domain corpora.