Knowledge-driven Natural Language Understanding of English Text and its Applications

Authors: Kinjal Basu, Sarat Chandra Varanasi, Farhad Shakerin, Joaquín Arias, Gopal Gupta12554-12563

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Table 1 summarizes the testing statistics and performance metrics of the SQuARE system (benchmarks are tested on an intel i9-9900 CPU with 16G RAM). Table 2 and table 3 compares our results in terms of accuracy with the existing state-of-the-art results for SQuARE and StaCACK system respectively.
Researcher Affiliation Academia 1 Department of Computer Science, University of Texas at Dallas, USA 2 Artificial Intelligence Research Group, Universidad Rey Juan Carlos, Madrid, Spain
Pseudocode Yes Algorithm 1 Semantic Knowledge Generation; Algorithm 2 Partial Tree Matching
Open Source Code No No explicit statement about the release of open-source code for the methodology described in this paper or a direct link to a code repository was found.
Open Datasets Yes The SQuARE and the StaCACK system have been tested on the bAbI QA (Weston et al. 2015) and the bAbI dialog dataset respectively (Bordes, Boureau, and Weston 2016).
Dataset Splits No The paper uses the bAbI QA and dialog datasets for evaluation but does not specify custom training, validation, and test splits for its own reasoning-based system, nor does it explicitly state using predefined splits of these datasets for reproduction in a machine learning context. It states: 'Our work is based purely on reasoning and and does not require any manual intervention other than providing (reusable) commonsense knowledge coded in ASP.'
Hardware Specification Yes Table 1 summarizes the testing statistics and performance metrics of the SQuARE system (benchmarks are tested on an intel i9-9900 CPU with 16G RAM).
Software Dependencies No The paper mentions using "Stanford Core NLP" and "spaCy" for parsing and "s(CASP) system" for ASP execution, but it does not specify exact version numbers for these software dependencies.
Experiment Setup No The paper describes the logical components and algorithms of its reasoning-based system but does not provide specific experimental setup details such as hyperparameters, batch sizes, or optimizer settings, as its approach is not based on machine learning training.