Asking without Telling: Exploring Latent Ontologies in Contextual Representations

Abstract

The success of pretrained contextual encoders, such as ELMo and BERT, has brought a great deal of interest in what these models learn: do they, without explicit supervision, learn to encode meaningful notions of linguistic structure? If so, how is this structure encoded? To investigate this, we introduce latent subclass learning (LSL): a modification to classifierbased probing that induces a latent categorization (or ontology) of the probe's inputs. Without access to fine-grained gold labels, LSL extracts emergent structure from input representations in an interpretable and quantifiable form. In experiments, we find strong evidence of familiar categories, such as a notion of personhood in ELMo, as well as novel ontological distinctions, such as a preference for fine-grained semantic roles on core arguments. Our results provide unique new evidence of emergent structure in pretrained encoders, including departures from existing annotations which are inaccessible to earlier methods.

Finding Universal Grammatical Relations in Multilingual BERT

“…Since syntactic subspaces are at most a small part of the total BERT space, these are not necessarily mutually contradictory with our results. In concurrent work, Michael et al (2020) also extend probing methodology, extracting latent ontologies from contextual representations without direct supervision.…”

A Cluster-based Approach for Improving Isotropy in Contextual Embedding Space

“…Hence, it might not be optimal for contextual embeddings, especially in the light that the latter tends to have a clustered structure. For instance, recent work suggests that word types (e.g., verbs, nouns, punctuations), entities (e.g., personhood, nationalities, and dates), and even word senses (Michael et al, 2020;Loureiro et al, 2021;Reif et al, 2019) create local distinct clustered areas in the contextual embedding space. Moreover, our local assessment shows that it is not necessarily the case that all clusters share the same dominant directions.…”

The Rediscovery Hypothesis: Language Models Need to Meet Linguistics

“…Meanwhile, methods were proposed that take into account not only the probing performance but also the ease of extracting linguistic information (Voita & Titov, 2020) or the complexity of the probing model (Pimentel et al, 2020a). At the same time, Wu et al (2020) and Michael et al (2020) suggested avoiding learnability issues by non-parametric probing 26 and weak supervision respectively. The remainder of the criticism is directed at the limitations of probing such as insufficient reliability for low-resourced languages (Eger et al, 2020), lack of evidence that probes indeed extract linguistic structures but do not learn from the linear context only (Kunz & Kuhlmann, 2020), lack of correlation with fine-tuning scores (Tamkin et al, 2020) and with pretraining scores (Ravichander et al, 2020;Elazar et al, 2021).…”