Abstract

To employ Pre-trained Language Models (PLMs) as knowledge containers in niche domains it is important to gauge the knowledge of these PLMs about facts in these domains. It is also an important pre-requisite to know how much enrichment effort is required to make them better. As part of this work, we aim to gauge and enrich small PLMs for knowledge of world geography. Firstly, we develop a moderately sized dataset of masked sentences covering 24 different fact types about world geography to estimate knowledge of PLMs on these facts. We hypothesize that for this niche domain, smaller PLMs may not be well equipped. Secondly, we enrich PLMs with this knowledge through fine-tuning and check if the knowledge in the dataset is infused sufficiently. We further hypothesize that linguistic variability in the manual templates used to embed the knowledge in masked sentences does not affect the knowledge infusion. Finally, we demonstrate the application of PLMs to tourism blog search and Wikidata KB augmentation. In both applications, we aim at showing the effectiveness of using PLMs to achieve competitive performance.

Abstract

Traditional evaluation metrics like BLEU and ROUGE fall short when capturing the nuanced qualities of generated text, particularly when there is no single ground truth. In this paper, we explore the potential of Large Language Models (LLMs), specifically Google Gemini 1, to serve as automatic evaluators for non-standardized metrics in summarization and dialog-based tasks. We conduct experiments across multiple prompting strategies to examine how LLMs fare as quality evaluators when compared with human judgments on the SummEval and USR datasets, asking the model to generate both a score as well as a justification for the score. Furthermore, we explore the robustness of the LLM evaluator by using perturbed inputs. Our findings suggest that while LLMs show promise, their alignment with human evaluators is limited, they are not robust against perturbations and significant improvements are required for their standalone use as reliable evaluators for subjective metrics.

Abstract

While documenting Architectural Knowledge (AK) is crucial, it is frequently neglected in many projects, and existing manual tools are underutilized. Although undocumented, Archi-tecture Knowledge (AK) is dispersed across various sources such as source code, documentation, and runtime logs. To address this, automated tools for efficient AK extraction and documentation are essential. Even after generating AK, navigating through vast the Architectural Records can be overwhelming. Building on that, we propose an automated Architectural Knowledge Management (AKM) System using Information Extraction and Generative AI, which generates AK from various source for a given system and answers architectural queries with respect to the given system. The development of an efficient Architectural Knowledge Management (AKM) system, which is both effective and user-friendly, entails the resolution of numerous challenges. It requires consolidating diverse AK data sources scattered across code, dia-grams, repository commits, and online platforms. The integration of Multimodal AI for AK extraction, incorporation of global AK, and leveraging Generative AI for AK documentation further compounds the problem. Moreover, generating contextually appropriate query responses adds another layer of complexity. To this end, we performed an initial exploratory study on generating Architectural Design Decisions using generative Large Language Models (LLM) in the context of Architecture Decision Records (ADR). Our initial results have been promising indicating the potential impact of GenAI for architectural knowledge management.

Abstract

Architectural Knowledge Management (AKM) involves the organized handling of information related to architectural decisions and design within a project or organization. An essential artefact of AKM is the Architecture Decision Records (ADR), which documents key design decisions. ADRs are documents that capture decision context, decision made and various aspects related to a design decision, thereby promoting transparency, collaboration, and understanding. Despite their benefits, ADR adoption in software development has been slow due to challenges like time constraints and inconsistent uptake. Recent advancements in Large Language Models (LLMs) may help bridge this adoption gap by facilitating ADR generation. However, the effectiveness of LLM for ADR generation or understanding is something that has not been explored. To this end, in this work, we perform an exploratory study which aims to investigate the feasibility of using LLM for the generation of ADRs given the decision context. In our exploratory study, we utilize GPT and T5-based models with 0-shot, few-shot, and fine-tuning approaches to generate the Decision of an ADR given its Context. Our results indicate that in a 0-shot setting, state-of-the-art models such as GPT-4 generate relevant and accurate Design Decisions, although they fall short of human-level performance. Additionally, we observe that more cost-effective models like GPT-3.5 can achieve similar outcomes in a few-shot setting, and smaller models such as Flan-T5 can yield comparable results after fine-tuning. To conclude, this exploratory study suggests that LLM can generate Design Decisions, but further research is required to attain human-level generation and establish standardized widespread adoption.

Abstract

This paper describes our approach for SemEval- 2024 Task 9: BRAINTEASER: A Novel Task Defying Common Sense. The BRAIN- TEASER task comprises multiple-choice Ques- tion Answering designed to evaluate the mod- els’ lateral thinking capabilities. It consists of Sentence Puzzle and Word Puzzle subtasks that require models to defy default common- sense associations and exhibit unconventional thinking. We propose a unique strategy to im- prove the performance of pre-trained language models, notably the Gemini 1.0 Pro Model, in both subtasks. We employ static and dy- namic few-shot prompting techniques and in- troduce a model-generated reasoning strategy that utilizes the LLM’s reasoning capabilities to improve performance. Our approach demon- strated significant improvements, showing that it performed better than the baseline models by a considerable margin but fell short of per- forming as well as the human annotators, thus highlighting the efficacy of the proposed strate- gies. We have made our code open-sourced for the replicability of our methods.

Abstract

Knowledge graphs (KGs) have been playing a crucial role in leveraging informa- tion on web for several downstream tasks, making it vital to construct and maintain them. Despite previous efforts in populating KGs, these methods typically do not focus on analyzing entity-specific content exclusively but rely on a fixed collection of documents. We define an approach to populate such KGs by utilizing entity-specific content on the web, for generating entity embeddings to establish entity-category interconnections. We empirically prove our ap- proach’s effectiveness, by utilizing it for a downstream task of Notability detection, associated with one of the most popular and important Knowledge Graphs - Wikipedia platform. To mod- erate the content uploaded to Wikipedia, “Notability” guidelines are defined by its editors to identify named entities that warrant their article on Wikipedia. So far notability is enforced by humans, which makes scalability an issue, and there has been no significant work on automat- ing this process. In this paper, we define a multipronged category-agnostic approach based on web-based entity features and their text-based salience encodings, to construct entity embed- dings for determining an entity’s notability. We distinguish entities based on their categories and utilize neural networks to perform classification. For validation, we utilize accuracy and prediction confidence on popular Wikipedia pages. Our system outperforms machine learning- based classifier approaches and handcrafted entity salience detection algorithms, by achieving performance accuracy of around 88%. Our system provides an efficient and scalable alterna- tive to manual decision-making about the importance of a topic, which could be extended to other such KG-based tasks

Abstract

Wikipedia is a highly essential platform because of its informative, dynamic, and easily accessible nature. To identify topics/titles warranting their own Wikipedia article, editors of Wikipedia defined “Notability” guidelines. So far notability is enforced by humans, which makes scalability an issue. There has been no significant work on Notability determination for titles with complex category dependencies. We design a mechanism to identify such titles. We construct a dataset with 9k such titles and propose a category-agnostic approach utilizing Graph neural networks, for their notability determination. Our system outperforms machine learning-based, transformer-based classifiers and entity salience methods. It provides a scalable alternative to manual decision-making about notability.

Abstract

Gauging the knowledge of Pretrained Language Models (PLMs) about facts in niche domains is an important step towards making them better in those domains. In this paper, we aim at evaluating multiple PLMs for their knowledge about world Geography. We contribute (i) a sufficiently sized dataset of masked Geography sentences to probe PLMs on masked token prediction and generation tasks, (ii) benchmark the performance of multiple PLMs on the dataset. We also provide a detailed analysis of the performance of the PLMs on different Geography facts.

Abstract

One crucial aspect of content organization is the creation of article outlines, which summarize the primary topics and subtopics covered in an article in a structured manner. This paper introduces a solution called XOutlineGen, which generates cross-lingual outlines for encyclopedic texts from reference articles. XOutlineGen uses the XWikiRef dataset, which consists of encyclopedic texts generated from reference articles and section titles. The dataset is enhanced with two new languages and three new domains, resulting in ∼92K articles. Our pipeline employs this dataset to train a two-step generation model, which takes the article title and set of references as inputs and produces the article outline.

Abstract

Multiple business scenarios require an automated generation of descriptive human-readable long text from structured input data, where the source is typically a high-resource language and the target is a low or medium resource language. We define the Cross-Lingual Fact to Long Text Generation (XFLT) as a novel natural language generation (NLG) task that involves generating descriptive and human-readable long text in a target language from structured input data (such as fact triples) in a source language. XFLT is challenging because of (a) hallucinatory nature of the state-of-the-art NLG models, (b) lack of good quality training data, and (c) lack of a suitable cross-lingual NLG metric. Unfortunately previous work focuses on different related problem settings (cross-lingual facts to short text or monolingual graph to text) and has made no efforts to handle hallucinations. In this paper, we contribute a novel dataset, XLALIGN with over 64,000 paragraphs across 12 different languages, and English facts. We propose a novel solution to the XFLT task which addresses these challenges by training multilingual Transformer-based encoder-decoder models with coverage prompts and grounded decoding. Further, it improves on the XFLT quality by defining task-specific reward functions and training on them using reinforcement learning. On XLALIGN, we compare this novel solution with several strong baselines using a new metric, cross-lingual PARENT. We also make our code and data publicly available