Farbod Zamani’s scientific contributions

What is this page?

This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (1)

Figure 1: Distribution of concreteness and sensory strength ratings for the 100 verb-noun polysemic phrases. Ratings (y axis) are normalized in the range 0-1. As shown by the averages (horizontal coloured lines), concrete phrases show higher concreteness and stronger involvement of all types of sensory information.
Figure 2: Pearson correlation between predicted and true variables for each model. We plot each crossvalidation split as a separate scatter point. XGLM consistently provides the best correlation scores across all variables.
Figure 3: Sense discrimination scores for each model, using all semantic variables. Error bars indicate the standard error of the mean across test splits. Overall indicate that the sense discrimination task is challenging for all models.
Figure 4: Overall sense discrimination scores for a number of contextualized models, across all layers. Overall, all versions of XGLM perform better in the first half of the layers. We indicate with a circle the layer used for the analyses reported above.
Figure 5: Pairwise similarities as measured by Representational Similarity Analysis among models. The scores reported in white are Pearson correlation scores, indicating a clear distinction between static and contextualized models.
Polysemy through the lens of psycholinguistic variables: a dataset and an evaluation of static and contextualized language models
  • Conference Paper
  • Full-text available

June 2024

·

62 Reads

·

Farbod Zamani

·

Polysemes are words that can have different senses depending on the context of utterance: for instance, 'newspaper' can refer to an organization (as in 'manage the newspaper') or to an object (as in 'open the newspaper'). Contrary to a large body of evidence coming from psy-cholinguistics, polysemy has been traditionally modelled in NLP by assuming that each sense should be given a separate representation in a lexicon (e.g. WordNet). This led to the current situation, where datasets used to evaluate the ability of computational models of semantics miss crucial details about the representation of polysemes, thus limiting the amount of evidence that can be gained from their use. In this paper we propose a framework to approach polysemy as a continuous variation in psycholinguistic properties of a word in context. This approach accommodates different sense interpretations, without postulating clear-cut jumps between senses. First we describe a publicly available English dataset that we collected , where polysemes in context (verb-noun phrases) are annotated for their concreteness and body sensory strength. Then, we evaluate static and contextualized language models in their ability to predict the ratings of each pol-yseme in context, as well as in their ability to capture the distinction among senses, revealing and characterizing in an interpretable way the models' flaws.

Download