English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Conditional text generation has been a challenging task that is yet to seehuman-level performance from state-of-the-art models. In this work, wespecifically focus on the Commongen benchmark, wherein the aim is to generate aplausible sentence for a given set of input concepts. Despite advances in othertasks, large pre-trained language models that are fine-tuned on this datasetoften produce sentences that are syntactically correct but qualitativelydeviate from a human understanding of common sense. Furthermore, generatedsequences are unable to fulfill such lexical requirements as matchingpart-of-speech and full concept coverage. In this paper, we explore howcommonsense knowledge graphs can enhance model performance, with respect tocommonsense reasoning and lexically-constrained decoding. We propose strategiesfor enhancing the semantic correctness of the generated text, which weaccomplish through: extracting commonsense relations from Conceptnet, injectingthese relations into the Unified Language Model (UniLM) through attentionmechanisms, and enforcing the aforementioned lexical requirements throughoutput constraints. By performing several ablations, we find that commonsenseinjection enables the generation of sentences that are more aligned with humanunderstanding, while remaining compliant with lexical requirements.

Lexically-constrained Text Generation through Commonsense Knowledge Extraction and Injection