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

Large Language Models (LLMs) have exhibited impressive generationcapabilities, but they suffer from hallucinations when solely relying on theirinternal knowledge, especially when answering questions that require lesscommonly known information. Retrieval-augmented LLMs have emerged as apotential solution to ground LLMs in external knowledge. Nonetheless, recentapproaches have primarily emphasized retrieval from unstructured text corpora,owing to its seamless integration into prompts. When using structured data suchas knowledge graphs, most methods simplify it into natural text, neglecting theunderlying structures. Moreover, a significant gap in the current landscape isthe absence of a realistic benchmark for evaluating the effectiveness ofgrounding LLMs on heterogeneous knowledge sources (e.g., knowledge base andtext). To fill this gap, we have curated a comprehensive dataset that poses twounique challenges: (1) Two-hop multi-source questions that require retrievinginformation from both open-domain structured and unstructured knowledgesources; retrieving information from structured knowledge sources is a criticalcomponent in correctly answering the questions. (2) The generation of symbolicqueries (e.g., SPARQL for Wikidata) is a key requirement, which adds anotherlayer of challenge. Our dataset is created using a combination of automaticgeneration through predefined reasoning chains and human annotation. We alsointroduce a novel approach that leverages multiple retrieval tools, includingtext passage retrieval and symbolic language-assisted retrieval. Our modeloutperforms previous approaches by a significant margin, demonstrating itseffectiveness in addressing the above-mentioned reasoning challenges.

DIVKNOWQA: Assessing the Reasoning Ability of LLMs via Open-Domain Question Answering over Knowledge Base and Text