Turing AI Fellowship: Event-Centric Framework for Natural Language Understanding
Lead Research Organisation:
University of Warwick
Department Name: Computer Science
Abstract
Natural language understanding (NLU) aims to allow computers to understand text automatically. NLU may seem easy to humans, but it is extremely difficult for computers because of the variety, ambiguity, subtlety, and expressiveness of human languages. Recent efforts to NLU have been largely exemplified in tasks such as natural language inference, reading comprehension and question answering. A common practice is to pre-train a language model such as BERT on large corpora to learn word representations and fine-tune on task-specific data. Although BERT and its successors have achieved state-of-the-art performance in many NLP tasks, it has been found that pre-trained language models mostly only reason about the surface form of entity names and fail to capture rich factual knowledge. Moreover, NLU models built on such pre-trained language models are susceptible to adversarial attack that even a small perturbation of an input (e.g., paraphrase questions and/or answers in QA tasks) would result in dramatic decrease in models' performance, showing that such models largely rely on shallow cues.
In human reading, successful reading comprehension depends on the construction of an event structure that represents what is happening in text, often referred to as the situation model in cognitive psychology. The situation model also involves the integration of prior knowledge with information presented in text for reasoning and inference. Fine-tuning pre-trained language models for reading comprehension does not help in building such effective cognitive models of text and comprehension suffers as a result.
In this fellowship, I aim to develop a knowledge-aware and event-centric framework for natural language understanding, in which event representations are learned from text with the incorporation of prior background and common-sense knowledge; event graphs are built on-the-fly as reading progresses; and the comprehension model is self-evolved to understand new information. I will primarily focus on reading comprehension and my goal is to enable computers to solve a variety of cognitive tasks that mimic human-like cognitive capabilities, bringing us a step closer to human-like intelligence.
In human reading, successful reading comprehension depends on the construction of an event structure that represents what is happening in text, often referred to as the situation model in cognitive psychology. The situation model also involves the integration of prior knowledge with information presented in text for reasoning and inference. Fine-tuning pre-trained language models for reading comprehension does not help in building such effective cognitive models of text and comprehension suffers as a result.
In this fellowship, I aim to develop a knowledge-aware and event-centric framework for natural language understanding, in which event representations are learned from text with the incorporation of prior background and common-sense knowledge; event graphs are built on-the-fly as reading progresses; and the comprehension model is self-evolved to understand new information. I will primarily focus on reading comprehension and my goal is to enable computers to solve a variety of cognitive tasks that mimic human-like cognitive capabilities, bringing us a step closer to human-like intelligence.
Publications
Adewoyin R
(2021)
TRU-NET: a deep learning approach to high resolution prediction of rainfall
in Machine Learning
Adewoyin R.A.
(2022)
RSTGen: Imbuing Fine-Grained Interpretable Control into Long-Form Text Generators
in NAACL 2022 - 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference
Arana-Catania M.
(2022)
Natural Language Inference with Self-Attention for Veracity Assessment of Pandemic Claims
in NAACL 2022 - 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference
Arana-Catania M.
(2021)
Evaluation of Abstractive Summarisation Models with Machine Translation in Deliberative Processes
in 3rd Workshop on New Frontiers in Summarization, NewSum 2021 - Workshop Proceedings
Dougrez-Lewis J
(2021)
Learning Disentangled Latent Topics for Twitter Rumour Veracity Classification
Dougrez-Lewis J.
(2022)
PHEMEPlus: Enriching Social Media Rumour Verification with External Evidence
in FEVER 2022 - 5th Fact Extraction and VERification Workshop, Proceedings of the Workshop
Fang Z
(2021)
A Query-Driven Topic Model
Related Projects
| Project Reference | Relationship | Related To | Start | End | Award Value |
|---|---|---|---|---|---|
| EP/V020579/1 | 01/01/2021 | 29/09/2022 | £1,269,626 | ||
| EP/V020579/2 | Transfer | EP/V020579/1 | 30/09/2022 | 30/03/2026 | £887,437 |
| Description | Our key findings are summarised below: (1) Event extraction - We have created a new dataset, PHEE, annotated with the adverse drug effect (ADE) events and the potential therapeutic effect (PTE) events, for training supervised models for the extraction of pharmacovigilance events from medical case reports. We have developed both extractive and generative Question-Answering (QA) approaches for extracting both ADE and PTE events from unstructured text in medical case reports. Apart from the health domain, we have also developed a novel approach to extraction financial events from documents in the finance domain. (2) Event semantic relation detection - We have developed novel approaches for event temporal relation extraction. The first approach was built on the hyperbolic geometry to capture the event-event asymmetric temporal relations, while the second approach was proposed to encode the event temporal relations from an external commonsense knowledge base in a Bayesian learning framework. (3) Event-centric QA - We have developed a series of approaches for Question-Answering (QA) in text which requires reasoning of event semantic relations in narratives, including (a) a novel QA model with contrastive learning and invertible event transformation; and (b) extractive and generative QA models with the event-related knowledge incorporated as constraints for model learning. (4) Model robustness and interpretability - We have proposed novel approaches for (a) addressing the token uniformity problem of the transformer architecture; (b) providing hierarchical interpretation of neural text classifiers; (c) generating explanations for recommender systems; (d) providing interpretation of predictive uncertainties of text classifiers built on pre-trained language models. |
| Exploitation Route | Since spoken and written communication plays a central part in our daily work and life, the proposed framework will have a profound impact on a variety of application areas, including drug discovery, intelligent virtual assistants, automated customer services, smart home, and question-answering in the finance and legal domains, benefiting industries such as healthcare, finance, law, insurance and education. |
| Sectors | Digital/Communication/Information Technologies (including Software) Education Financial Services and Management Consultancy Healthcare Pharmaceuticals and Medical Biotechnology |
| URL | https://sites.google.com/view/yulanhe/narrative-understanding |
| Description | We have developed a range of web tools based on our research, including: (1) NarrativePlay: a novel system that allows users to role-play a fictional character and interact with other characters in narratives in an immersive environment. The system leverages LLMs to generate human-like responses, guided by personality traits extracted from narratives. The system incorporates auto-generated visual display of narrative settings, character portraits, and character speech, greatly enhancing the user experience. (2) DrugWatch: an easy-to-use and interactive multi-source information visualisation platform for drug safety study. It allows users to understand common side effects of drugs and their statistical information, flexibly retrieve relevant medical reports, or annotate their own medical texts with our automated annotation tool. (3) AERA Chat: an interactive platform which provides visually explained assessment of student answers and streamlines the verification of rationales. Users can input questions and student answers to obtain automated, explainable assessment results from LLMs. The platform's innovative visualization features and robust evaluation tools make it useful for educators to assist their marking process, and for researchers to evaluate assessment performance and quality of rationales generated by different LLMs, or as a tool for efficient annotation. |
| First Year Of Impact | 2024 |
| Sector | Digital/Communication/Information Technologies (including Software),Education,Pharmaceuticals and Medical Biotechnology |
| Impact Types | Societal Economic Policy & public services |
| Description | Elandi: Trustworthy generative AI for affordable personalised L&D |
| Amount | £799,724 (GBP) |
| Funding ID | 10093055 |
| Organisation | Innovate UK |
| Sector | Public |
| Country | United Kingdom |
| Start | 04/2024 |
| End | 04/2025 |
| Title | A QA model with contrastive learning and invertible event transformation (TranCLR) |
| Description | Human reading comprehension often requires reasoning of event semantic relations in narratives, represented by Event-centric Question-Answering (QA). To address event-centric QA, we propose a novel QA model with contrastive learning and invertible event transformation, call TranCLR. Our proposed model utilizes an invertible transformation matrix to project semantic vectors of events into a common event embedding space, trained with contrastive learning, and thus naturally inject event semantic knowledge into mainstream QA pipelines. The transformation matrix is fine-tuned with the annotated event relation types between events that occurred in questions and those in answers, using event-aware question vectors. Experimental results on the Event Semantic Relation Reasoning (ESTER) dataset show significant improvements in both generative and extractive settings compared to the existing strong baselines, achieving over 8.4% gain in the token-level F1 score and 3.0% gain in Exact Match (EM) score under the multi-answer setting. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | The model is described in a paper published in the Findings of EMNLP 2022. |
| URL | https://github.com/LuJunru/TranCLR |
| Title | Addressing token uniformity in transformers using the singular value transformation function (SoftDecay) |
| Description | Token uniformity is commonly observed in transformer-based models, in which different tokens share a large proportion of similar information after going through stacked multiple self-attention layers in a transformer. We propose to use the distribution of singular values of outputs of each transformer layer to characterise the phenomenon of token uniformity and empirically illustrate that a less skewed singular value distribution can alleviate the token uniformity problem. Base on our observations, we define several desirable properties of singular value distributions and propose a novel transformation function for updating the singular values. We show that apart from alleviating token uniformity, the transformation function should preserve the local neighbourhood structure in the original embedding space. Our proposed singular value transformation function is applied to a range of transformer-based language models such as BERT, ALBERT, RoBERTa and DistilBERT, and improved performance is observed in semantic textual similarity evaluation and a range of GLUE tasks |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | The proposed approach is described in a paper published in UAI 2022. |
| URL | https://github.com/hanqi-qi/tokenUni |
| Title | BERT-based Text and Image multimodal model with Contrasting learning (BTIC) |
| Description | The BERT-based Text and Image multimodal model with Contrasting learning (BTIC) has been developed for unreliable multimodal news detection. It captures both textual and visual information from unreliable articles utilising the contrastive learning strategy. The contrastive learner interacts with the unreliable news classifier to push similar credible news (or similar unreliable news) closer while moving news articles with similar content but opposite credibility labels away from each other in the multimodal embedding space. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | NA |
| URL | https://github.com/WenjiaZh/BTIC |
| Title | Bayesian Learning-based Translation (Bayesian-Trans) model for Event TempRel Extraction |
| Description | Existing models to extract temporal relations between events lack a principled method to incorporate external knowledge.We propose Bayesian-Trans, a Bayesian learning-based method that models the temporal relation representations as latent variables and infers their values via Bayesian inference and translational functions. Compared to conventional neural approaches, instead of performing point estimation to find the best set parameters, the proposed model infers the parameters' posterior distribution directly, enhancing the model's capability to encode and express uncertainty about the predictions. Experimental results on the three widely used datasets show that Bayesian-Trans outperforms existing approaches for event temporal relation extraction. We additionally present detailed analyses on uncertainty quantification, comparison of priors, and ablation studies, illustrating the benefits of the proposed approach. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The approach is described in a paper accepted to EACL 2023. |
| URL | https://github.com/Xingwei-Warwick/Bayesian-Trans |
| Title | CMM (Construction- Modelling-Modification): a scalable framework for Table-of-Contents extraction from ESG reports |
| Description | CMM is a new framework for Table-of-Contents (ToC) extraction, consisting of three steps: (1) Constructing an initial tree of text blocks based on reading order and font sizes; (2) Modelling each tree node (or text block) independently by considering its con- textual information captured in node-centric subtree; (3) Modifying the original tree by taking appropriate action on each tree node (Keep, Delete, or Move). This construction-modelling- modification (CMM) process offers several benefits. It eliminates the need for pairwise modelling of section headings as in previous approaches, making document segmentation practically feasible. By incorporating structured information, each section heading can leverage both local and long-distance context relevant to itself. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The approach is presented in a paper published in EMNLP 2023 (https://aclanthology.org/2023.emnlp-main.816.pdf). |
| URL | https://github.com/xnyuwg/cmm |
| Title | CUE: a text Classifier Uncertainty Explanation model |
| Description | CUE aims to interpret uncertainties inherent in the predictions of text classifiers built on Pre-trained Language Models (PLMs). In particular, we first map PLM-encoded representations to a latent space via a variational auto-encoder. We then generate text representations by perturbing the latent space which causes fluctuation in predictive uncertainty. By comparing the difference in predictive uncertainty between the perturbed and the original text representations, we are able to identify the latent dimensions responsible for uncertainty and subsequently trace back to the input features that contribute to such uncertainty. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The model is proposed in a paper published in UAI 2023. |
| URL | https://github.com/lijiazheng99/CUE |
| Title | Conan: a benchmark for extracting and analysing intricate character relation graphs from detective narratives |
| Description | The creation of Conan is for assessing the capabilities of LLMs to comprehend nuanced relational dynamics in narrative contexts. Conan is designed for extracting and analysing intricate character relation graphs from detective narratives. We designed hierarchical relationship categories and manually extracted and annotated role-oriented relationships from the perspectives of various characters, incorporating both public relationships known to most characters and secret ones known to only a few. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| Impact | Conan was just released in February 2024. |
| URL | https://github.com/BLPXSPG/Conan |
| Title | Cross-modal PROtotype driven NETwork (XPRONET) |
| Description | Radiology report generation (RRG) aims to describe automatically a radiology image with human-like language. As an alternative to expert diagnosis, RRG could potentially support the work of radiologists, reducing the burden of manual reporting. Previous approaches often adopt an encoder-decoder architecture and focus on single-modal feature learning, while few studies explore cross-modal feature interaction. We propose a Cross-modal PROtotype driven NETwork (XPRONET) to promote cross-modal pattern learning and exploit it to improve the task of radiology report generation. This is achieved by three well-designed, fully differentiable and complementary modules: a shared cross-modal prototype matrix to record the cross-modal proto- types; a cross-modal prototype network to learn the cross-modal prototypes and embed the cross-modal information into the visual and textual features; and an improved multi-label contrastive loss to enable and enhance multi-label prototype learning. Experimental results demonstrate that XPRONET obtains substantial improvements on two commonly used medical report generation benchmark datasets, i.e., IU-Xray and MIMIC-CXR, where its performance exceeds recent state-of-the-art approaches by a large margin on IU-Xray dataset and achieves the SOTA performance on MIMIC-CXR. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | The model is described in a paper published in ECCV 2022. It has 3 paper citation, 22 stars, 3 forks. |
| URL | https://github.com/Markin-Wang/XProNet |
| Title | DIVA - the Disentangling Interaction of VAriables framework proposed for causal inference from text |
| Description | Adjusting for latent covariates is crucial for estimating causal effects from observational textual data. Most existing methods only account for confounding covariates that affect both treatment and outcome, potentially leading to biased causal effects. This bias arises from insufficient consideration of non-confounding covariates, which are relevant only to either the treatment or the outcome. Our proposed framework DIVA can mitigate the bias by unveiling interactions between different variables to disentangle the non-confounding covariates when estimating causal effects from text. The disentangling process ensures covariates only contribute to their respective objectives, enabling independence between variables. Additionally, we impose a constraint to balance representations from the treatment group and control group to alleviate selection bias. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The approach is presented in a paper accepted to the Findings of EMNLP 2023 (https://aclanthology.org/2023.findings-emnlp.709.pdf). |
| URL | https://github.com/zyxnlp/DIVA |
| Title | DOC -- Disentangled Opinion Clustering |
| Description | The Disentangled Opinion Clustering (DOC) model is proposed for vaccination opinion mining from social media. DOC is able to disentangle users' stances from opinions via a disentangling attention mechanism and a Swapping-Autoencoder, and is designed to process unseen aspect categories via a clustering approach, leveraging clustering-friendly representations induced by out-of-the-box Sentence-BERT encodings and disentangling mechanisms. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The approach is described in a paper published in Findings of ACL 2023 (https://aclanthology.org/2023.findings-acl.115.pdf). |
| URL | https://github.com/somethingx1202/DOC |
| Title | Dynamic Brand-Topic Model (dBTM) |
| Description | Monitoring online customer reviews is important for business organisations to measure customer satisfaction and better manage their reputations. We propose a novel dynamic Brand-Topic Model (dBTM) which is able to automatically detect and track brand-associated sentiment scores and polarity-bearing topics from product reviews organised in temporally-ordered time intervals. dBTM models the evolution of the latent brand polarity scores and the topic-word distributions over time by Gaussian state space models. It also incorporates a meta learning strategy to control the update of the topic-word distribution in each time interval in order to ensure smooth topic transitions and better brand score predictions. It has been evaluated on a dataset constructed from MakeupAlley reviews and a hotel review dataset. Experimental results show that dBTM outperforms a number of competitive baselines in brand ranking, achieving a good balance of topic coherence and uniqueness, and extracting well-separated polarity-bearing topics across time intervals. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | The model is described in a paper accepted by the Transactions of the Association for Computational Linguistics (TACL). |
| URL | https://github.com/BLPXSPG/dBTM |
| Title | ESGDoc - a new dataset designed for Table of Contents extraction from ESG reports |
| Description | An ESG dataset comprises 1,093 publicly available ESG annual reports, sourced from 563 distinct companies, and spans the period from 2001 to 2022. The reports vary in length, ranging from 4 pages to 521 pages, with an average of 72 pages. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The dataset is presented in a paper published in EMNLP 2023 (https://aclanthology.org/2023.emnlp-main.816.pdf). |
| URL | https://github.com/xnyuwg/cmm |
| Title | Event temporal relation extraction via hyperbolic geometry |
| Description | Detecting events and their evolution through time is a crucial task in natural language understanding. Recent neural approaches to event temporal relation extraction typically map events to embeddings in the Euclidean space and train a classifier to detect temporal relations between event pairs. However, embeddings in the Euclidean space cannot capture richer asymmetric relations such as event temporal relations. We thus propose to embed events into hyperbolic spaces, which are intrinsically oriented at modeling hierarchical structures. We introduce two approaches to encode events and their temporal relations in hyperbolic spaces. One approach leverages hyperbolic embeddings to directly infer event relations through simple geometrical operations. In the second one, we devise an end-to-end architecture composed of hyperbolic neural units tailored for the temporal relation extraction task. Thorough experimental assessments on widely used datasets have shown the benefits of revisiting the tasks on a different geo- metrical space, resulting in state-of-the-art performance on several standard metrics. Finally, the ablation study and several qualitative analyses highlighted the rich event semantics implicitly encoded into hyperbolic space. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | The proposed approaches are described in a paper published in EMNLP 2022. It has attracted 11 paper citation, 9 stars, 1 fork. |
| URL | https://github.com/Xingwei-Warwick/hyper-event-TempRel |
| Title | Explainable Automated Student Answer Assessment |
| Description | We introduce a novel framework that explores using ChatGPT, a cutting-edge large language model, for the concurrent tasks of student answer scoring and rationale generation. We identify the appropriate instructions by prompting ChatGPT with different templates to collect the rationales, where inconsistent rationales are refined to align with marking standards. The refined ChatGPT outputs enable us to fine-tune a smaller language model that simultaneously assesses student answers and provides rationales. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The model was described in a paper published in Findings of EMNLP 2023. |
| URL | https://github.com/lijiazheng99/aera |
| Title | Hierarchical Interpretable Neural Text classifier (HINT) |
| Description | Recent years have witnessed increasing interest in developing interpretable models in Natural Language Processing (NLP). Most existing models aim at identifying input features such as words or phrases important for model predictions. Neural models developed in NLP, however, often compose word semantics in a hierarchical manner. As such, interpretation by words or phrases only cannot faithfully explain model decisions in text classification.We propose a novel Hierarchical Interpretable Neural Text classifier, called HINT, which can automatically generate explanations of model predictions in the form of label-associated topics in a hierarchical manner. Model interpretation is no longer at the word level, but built on topics as the basic semantic unit. Experimental results on both review datasets and news datasets show that our proposed approach achieves text classification results on par with existing state-of-the-art text classifiers, and generates interpretations more faithful to model predictions and better understood by humans than other interpretable neural text classifiers |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | The approach is described in a paper published in the Computational Linguistics journal. |
| URL | https://github.com/hanqi-qi/HINT |
| Title | InfoAC: Information-Augmented and Consistency-Enhanced approach |
| Description | An Information-Augmented and Consistency-Enhanced fine-tuning approach to alleviate the sensitivity of causal language models to the order of in-context examples. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| Impact | Received 6 citations. |
| URL | https://github.com/xyzCS/InfoAC |
| Title | Joint Aspect-Sentiment Extraction (JASE) model for understanding patient reviews |
| Description | The Joint Aspect-Sentiment Extraction (JASE) model is developed for simultaneous unsupervised aspect extraction at the sentence level and supervised sentiment classification at the document level. It can extract coherent aspects from patient reviews and can automatically infer the distribution of aspects under different polarities without requiring aspect-level annotations for model learning. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | The model achieves 98.2% sentiment classification accuracy when tested on the reviews about healthcare services collected from Yelp, outperforming several strong baselines. |
| URL | https://github.com/GuiLinNLP/PatientReviewUnderstanding |
| Title | Knowledge-Aware Graph (KAG) Model for Emotion Cause Extraction |
| Description | Knowledge-Aware Graph (KAG) Model aims to identify the triggering factors for emotions expressed in text. It explicitly models the emotion triggering paths by leveraging the commonsense knowledge to enhance the semantic dependencies between a candidate clause and an emotion clause. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | The model has already extracted interests from industry such as WeChat AI from Tencent. |
| URL | https://github.com/hanqi-qi/Position-Bias-Mitigation-in-Emotion-Cause-Analysis |
| Title | MATTE -- a doMain AdapTive counTerfactual gEneration model |
| Description | Counterfactual generation lies at the core of various machine learning tasks. Existing disentangled methods crucially rely on oversimplified assumptions, such as assuming independent content and style variables, to identify the latent variables, even though such assumptions may not hold for complex data distributions. This problem is exacerbated when data are sampled from multiple domains since the dependence between content and style may vary significantly over domains. We proposed the doMain AdapTive counTerfactual gEneration model, called (MATTE), which addresses the domain-varying dependence between the content and the style variables inherent in the counterfactual generation task. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The model is presented in a paper published in NeurIPS 2023 (https://openreview.net/pdf?id=cslnCXE9XA). |
| URL | https://github.com/hanqi-qi/Matte |
| Title | MIRROR: Multiple-perspective self-reflection method for knowledge-rich reasoning |
| Description | While Large language models (LLMs) have the capability to iteratively reflect on their own outputs, recent studies have observed their struggles with knowledge-rich problems without access to external resources. In addition to the inefficiency of LLMs in self-assessment, we also observe that LLMs struggle to revisit their predictions despite receiving explicit negative feedback. Therefore, we proposed Mirror, a Multiple-perspective self-reflection method for knowledge-rich reasoning, to avoid getting stuck at a particular reflection iteration. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| Impact | 6 paper citations, 12 Github stars |
| URL | https://github.com/hanqi-qi/Mirror |
| Title | MOBO: The MOvie and BOok reviews dataset |
| Description | The MOBO dataset. The MOvie and BOok reviews dataset is a collection made up of movie and book reviews, paired with their related plots.
The reviews come from different publicly available datasets: the Stanford's IMDB movie reviews [1], the GoodReads [2] and the Amazon reviews dataset [3]. With the help of 15 annotators, we further labeled more than 18,000 reviews' sentences (~6000 per corpus), marking the sentence polarity (Positive, Negative), or whether a sentence describes its corresponding movie/book Plot, or none of the above (None). In the
dataset folder, we have shared an excerpt of the annotated sentences for each dataset. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | Until March 2024, the dataset has received 177 downloads. |
| URL | https://zenodo.org/record/6348893 |
| Title | MOBO: The MOvie and BOok reviews dataset |
| Description | The MOvie and BOok reviews dataset is a collection made up of movie and book reviews, paired with their related plots. The reviews come from different publicly available datasets: the Stanford's IMDB movie reviews, the GoodReads and the Amazon reviews dataset. With the help of 15 annotators, we further labeled more than 18,000 reviews' sentences (~6000 per corpus), marking the sentence polarity (Positive, Negative), or whether a sentence describes its corresponding movie/book plot, or none of the above (None). |
| Type Of Material | Database/Collection of data |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | Since the dataset was published in 2021, it has been cited by authors from Baidu Research in the US, the Institute for Research in Biomedicine (IRB) in Spain, the Universitat Politècnica de València in Spain, and the University of Sao Paulo in Brazil. |
| URL | https://zenodo.org/record/6348894#.Yix8pBDP1f0 |
| Title | MemoChat: Tuning LLMs to Use Memos for Consistent Long-Range Open-domain Conversation |
| Description | MemoChat enables large language models (LLMs) to effectively employ self-composed memos for maintaining consistent long-range conversations. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The model is described in a paper currently under review. |
| URL | https://github.com/LuJunru/MemoChat |
| Title | NapSS: Paragraph-level Medical Text Simplification via Narrative Prompting and Sentence-matching Summarization |
| Description | Accessing medical literature is difficult for laypeople as the content is written for specialists and contains medical jargon. Automated text simplification methods offer a potential means to address this issue. We propose a summarize-then-simplify two-stage strategy, which we call NapSS, identifying the relevant content to simplify while ensuring that the original narrative flow is preserved. In this approach, we first generate reference summaries via sentence matching between the original and the simplified abstracts. These summaries are then used to train an extractive summarizer, learning the most relevant content to be simplified. Then, to ensure the narrative consistency of the simplified text, we synthesize auxiliary narrative prompts combining key phrases derived from the syntactical analyses of the original text. Our model achieves results significantly better than the seq2seq baseline on an English medical corpus, yielding 3%~4% absolute improvements in terms of lexical similarity, and providing a further 1.1% improvement of SARI score when combined with the baseline. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | The approach is described in a paper accepted to the Findings of EACL 2023. |
| URL | https://github.com/LuJunru/NapSS |
| Title | OpenToM: a new benchmark for assessing LLMs' Neural Theory-of-Mind capabilities |
| Description | The OpenToM benchmark contains 696 narratives, 596 of which are narratives of normal length (average word count: 194.3 words) and 100 of which are long narratives (average word count: 491.6 words). Each of the narrative is followed with 23 ToM questions, making a total of 16,008 questions. The OpenToM benchmark pose first-order and second-order questions in the following genres: -- Location: this is a prevalent type of question seen in many ToM benchmarks. We break location questions into coarse and fine, differ by granularity. Coarse questions ask if a character thinks that an entity is in its initial location where as fine questions ask the precise location of an entity. -- Multihop: we compose questions that demand an additional reasoning hop on top of the Location questions. Specifically, we inquire characters' perception of the fullness and the accessibility of an entity. We incorporate social commonsense in the accessibility questions. For instance, if an entity is moved into someone's bag, then it becomes less accessible to others since people shall not access other's bag without asking for permission. -- Attitude: LLMs' capability of understanding character's perception of the psychological world has been overlooked by many established N-ToM benchmarks. We propose the attitude question to test LLMs' capabilities in understanding character's attitude towards some events. For instance, if my favourite rubber duck is taken away from me without asking, I would hold a negative attitude towards this event. All the OpenToM questions are designed to be a binary or ternary classification task. We recommend using macro-averaged F1 score to evaluate LLMs' performance as the labels are not uniformly distributed. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| Impact | The dataset is released on Hugging Face. |
| URL | https://huggingface.co/datasets/SeacowX/OpenToM |
| Title | PHEE: A Dataset for Pharmacovigilance Event Extraction from Text |
| Description | The PHEE dataset contains over 5,000 finely annotated pharmacovigilance events from public medical case reports. Two types of events, the adverse events and the potential therapeutic events, are annotated. For each event, we annotate the event trigger and hierarchical arguments. The main arguments (coarse-grained spans) include subject, treatment and effect. Further fine-grained sub-arguments - age, gender, race, number of patients (labelled as population) and preexisting conditions (labelled as subject.disorder) for the subject argument and drug (and their combinations), dosage, frequency, route, time-elapsed, duration, target disorder (labelled as treatment.disorder) for the treatment argument - are then annotated upon main arguments. We provide two formats of data: visualisation-friendly brat-format data and structured json data for the convenience of use. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | The dataset is described in a paper published in EMNLP 2022. |
| URL | https://zenodo.org/record/7689970 |
| Title | ProCNet: Proxy Nodes Clustering Network for document-level event extraction |
| Description | ProCNet is a model proposed for document-level multi-event extraction with event proxy nodes and Hausdorff distance minimisation. The event proxy nodes, representing pseudo-events, are able to build connections with other event proxy nodes, essentially capturing global information. The Hausdorff distance makes it possible to compare the similarity between the set of predicted events and the set of ground-truth events. By directly minimising Hausdorff distance, the model is trained towards the global optimum directly, which improves performance and reduces training time. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2023 |
| Provided To Others? | Yes |
| Impact | ProCNet is described in a paper published in ACL 2023 (https://aclanthology.org/2023.acl-long.563.pdf) |
| URL | https://github.com/xnyuwg/procnet |
| Title | Prompt preference dataset |
| Description | The prompt preference dataset is constructed to facilitate the training of our proposed Free-form Instruction-oriented Prompt Optimisation (FIPO) framework. It contains about 30k instructions and is constructed using both the optimal and suboptimal LLMs, undergoing rigorous cross-validation by human experts and analytical models. The FIPO framework reconceptualises the prompt optimisation process into manageable modules, anchored by a meta prompt that dynamically adjusts its content. This allows for the flexible integration of the task instructions, optional instruction responses, and optional ground truth to produce finely optimised task prompts. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| Impact | The dataset has been uploaded to Hugging Face in February 2024. |
| URL | https://huggingface.co/datasets/Junrulu/Prompt_Preference_Dataset |
| Title | Stance-Augmented VAE Disentanglement model (SAVED) |
| Description | The SAVED model has been proposed for Twitter rumour veracity assessment. It incorporates a Variational Auto Encoder (VAE) with adversarial learning to disentangle topics which are informative for stance classification from those which are not. Tweet representations are derived based on the word representations learned in the latent stance-dependent topic space, which are then used to train a veracity classifier to classify whether the veracity of an input tweet is true, false or unverified. The model achieves the state-of-the-art accuracy scores on the commonly used PHEME dataset for Twitter veracity assessment. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | The developed SAVED model achieves the state-of-the-art accuracy scores on the commonly used PHEME dataset for Twitter veracity assessment. |
| URL | https://github.com/JohnNLP/SAVED |
| Title | Stance-Aware Evidence Reasoning and Stance-Aware Aggregation model (TARSA) |
| Description | TARSA was proposed for more accurate fact verification, with the following four key properties: 1) checking topical consistency between the claim and evidence; 2) maintaining topical coherence among multiple pieces of evidence; 3) ensuring semantic similarity between the global topic information and the semantic representation of evidence; 4) aggregating evidence based on their implicit stances to the claim. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2021 |
| Provided To Others? | Yes |
| Impact | Since the paper was published in 2021, it has been cited by authors from Checkstep Research, Sofia University in Bulgaria, University of Copenhagen in Denmark, Qatar Computing Research Institute, Fudan University in China, ByteDance AI Lab and the University of California, Santa Barbara in the US. |
| URL | https://github.com/jasenchn/TARSA |
| Title | Vaccine Attitude Detection (VAD) model |
| Description | We propose a novel semi-supervised approach for vaccine attitude detection, called VADET. A variational autoencoding architecture based on language models is employed to learn from unlabelled data the topical information of the domain. Then, the model is fine-tuned with a few manually annotated examples of user attitudes. We validate the effectiveness of VADET on our annotated data and also on an existing vaccination corpus annotated with opinions on vaccines. Our results show that VADET is able to learn disentangled stance and aspect topics, and outperforms existing aspect-based sentiment analysis models on both stance de- tection and tweet clustering. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | The model is proposed in a paper published in NAACL 2022. |
| URL | https://github.com/somethingx1202/VADet |
| Description | Collaboration with AstraZeneca |
| Organisation | AstraZeneca |
| Country | United Kingdom |
| Sector | Private |
| PI Contribution | We have created an annotated text corpus for adverse drug event (ADE) extraction for pharmacovigilance study, and have developed the QA approaches for the extraction of ADEs, and their associated subjects, treatments, and effects. Further more, we have also developed approaches for more fine-grained extract by identifying patient demographic information, the context information about the treatments such as drug dosage levels, administration routes, frequency, and attributes relating to the events. |
| Collaborator Contribution | Colleagues from AstraZeneca participate regular meetings and provide suggestions and feedback on our research work. |
| Impact | https://aclanthology.org/2022.emnlp-main.376/ |
| Start Year | 2021 |
| Description | Collaboration with Northeastern University, US |
| Organisation | Northeastern University - Boston |
| Country | United States |
| Sector | Academic/University |
| PI Contribution | We have created an annotated text corpus for adverse drug event (ADE) extraction for pharmacovigilance study, and have developed the QA approaches for the extraction of ADEs, and their associated subjects, treatments, and effects. Further more, we have also developed approaches for more fine-grained extract by identifying patient demographic information, the context information about the treatments such as drug dosage levels, administration routes, frequency, and attributes relating to the events. |
| Collaborator Contribution | Given the prior experience of creating an annotated dataset for Randomised Control Trials by our collaborator, Prof. Byron Wallace, we had discussions to define the annotation schema, guidelines and the annotation procedure. Prof. Wallace attends our biweekly project meetings and provides feedback and suggestions on our pharmacovigilance event extraction research. |
| Impact | https://aclanthology.org/2022.emnlp-main.376/ https://aclanthology.org/2024.eacl-short.30/ |
| Start Year | 2022 |
| Title | A Disentangled Adversarial Neural Topic Model for Separating Opinions from Plots in User Reviews |
| Description | This is the code of the DIATOM model presented in the NAACL 2021 paper: A Disentangled Adversarial Neural Topic Model for Separating Opinions from Plots in User Reviews, G. Pergola, L. Gui, Y. He, NAACL 2021 [link] Abstract: "The flexibility of the inference process in Variational Autoencoders (VAEs) has recently led to revising traditional probabilistic topic models giving rise to Neural Topic Models (NTM). Although these approaches have achieved significant results, surprisingly very little work has been done on how to disentangle the latent topics. Existing topic models when applied to reviews may extract topics associated with writers' subjective opinions mixed with those related to factual descriptions such as plot summaries in movie and book reviews. It is thus desirable to automatically separate opinion topics from plot/neutral ones enabling a better interpretability. In this paper, we propose a neural topic model combined with adversarial training to disentangle opinion topics from plot and neutral ones. We conduct an extensive experimental assessment introducing a new collection of movie and book reviews paired with their plots, namely MOBO dataset, showing an improved coherence and variety of topics, a consistent disentanglement rate, and sentiment classification performance superior to other supervised topic models." |
| Type Of Technology | Software |
| Year Produced | 2022 |
| Open Source License? | Yes |
| URL | https://zenodo.org/record/6349198 |
| Title | A Neural Generative Model for Joint Learning Topics and Topic-Specific Word Embeddings |
| Description | topical_wordvec_models You first need to create a save folder for training. Download the [saved model](https://topicvecmodels.s3.eu-west-2.amazonaws.com/save/47/model) and place it in ./save/47/ to run the trained model. To construct the training set, refer to https://github.com/somethingx02/topical_wordvec_model please. Trained [wordvecs](https://topicvecmodels.s3.eu-west-2.amazonaws.com/save/47/aggrd_all_wordrep.txt). |
| Type Of Technology | Software |
| Year Produced | 2022 |
| Open Source License? | Yes |
| URL | https://zenodo.org/record/6352449 |
| Title | A Neural Generative Model for Joint Learning Topics and Topic-Specific Word Embeddings |
| Description | topical_wordvec_models You first need to create a save folder for training. Download the [saved model](https://topicvecmodels.s3.eu-west-2.amazonaws.com/save/47/model) and place it in ./save/47/ to run the trained model. To construct the training set, refer to https://github.com/somethingx02/topical_wordvec_model please. Trained [wordvecs](https://topicvecmodels.s3.eu-west-2.amazonaws.com/save/47/aggrd_all_wordrep.txt). |
| Type Of Technology | Software |
| Year Produced | 2022 |
| Open Source License? | Yes |
| URL | https://zenodo.org/record/6352450 |
| Title | CHIME: Cross-passage Hierarchical Memory Network for Generative Review Question Answering |
| Description | We introduce CHIME, a cross-passage hierarchical memory network for question answering (QA) via text generation. It extends XLNet introducing an auxiliary memory module consisting of two components: the context memory collecting cross-passage evidences, and the answer memory working as a buffer continually refining the generated answers. Empirically, we show the efficacy of the proposed architecture in the multi-passage generative QA, outperforming the state-of-the-art baselines with better syntactically well-formed answers and increased precision in addressing the questions of the AmazonQA review dataset. An additional qualitative analysis revealed the interpretability introduced by the memory module. |
| Type Of Technology | Software |
| Year Produced | 2020 |
| Open Source License? | Yes |
| Impact | CHIME is a cross-passage hierarchical memory network for generative question answering (QA). It extends XLNet introducing an auxiliary memory module consisting of two components: the context memory collecting cross-passage evidences, and the answer memory working as a buffer continually refining the generated answers. A sample of syntactically well-formed answers show the efficacy of CHIME. This repository contains PyTorch implementation of the corresponding COLING 2020 Paper. |
| URL | https://zenodo.org/record/6351416 |
| Title | Code for CIKM Short Paper "Supervised Contrastive Learning for Multimodal Unreliable News Detection in COVID-19 Pandemic" |
| Description | The code for CIKM short paper "Supervised Contrastive Learning for Multimodal Unreliable News Detection in COVID-19 Pandemic". In this work, we propose a BERT-based multimodal unreliable news detection framework, which captures both textual and visual information from unreliable articles utilising the contrastive learning strategy. The contrastive learner interacts with the unreliable news classifier to push similar credible news (or similar unreliable news) closer while moving news articles with similar content but opposite credibility labels away from each other in the multimodal embedding space. |
| Type Of Technology | Software |
| Year Produced | 2021 |
| Open Source License? | Yes |
| Impact | NA |
| URL | https://zenodo.org/record/6342230 |
| Title | Code for EMNLP paper "Extracting Event Temporal Relations via Hyperbolic Geometry" |
| Description | This is the code of EMNLP 2021 main track long paper "Extracting Event Temporal Relations via Hyperbolic Geometry". The paper proposed two hyperbolic-based approaches for the event temporal relation extraction task, which is an Event-centric Natural Language Understanding task. |
| Type Of Technology | Software |
| Year Produced | 2022 |
| Open Source License? | Yes |
| Impact | Since its recent release, the software available on GitHub has already been starred several times. |
| URL | https://zenodo.org/record/6349213 |
| Title | Code for EMNLP paper "Extracting Event Temporal Relations via Hyperbolic Geometry" |
| Description | This is the code of EMNLP 2021 main track long paper "Extracting Event Temporal Relations via Hyperbolic Geometry". The paper proposed two hyperbolic-based approaches for the event temporal relation extraction task, which is an Event-centric Natural Language Understanding task. |
| Type Of Technology | Software |
| Year Produced | 2022 |
| Open Source License? | Yes |
| URL | https://zenodo.org/record/6349212 |
| Title | DIATOM - A Disentangled Adversarial Neural Topic Model |
| Description | Implementation of the DIATOM model presented in the NAACL 2021 paper: "A Disentangled Adversarial Neural Topic Model for Separating Opinions from Plots in User Reviews", G. Pergola, L. Gui, Y. He, NAACL 2021 |
| Type Of Technology | Software |
| Year Produced | 2021 |
| Open Source License? | Yes |
| Impact | The software available on GitHub has already been starred several times. |
| Title | DrugWatch: A Comprehensive Multi-Source Data Visualisation Platform for Drug Safety Information |
| Description | DrugWatch is an easy-to-use and interactive multi-source information visualisation platform for drug safety study. It allows users to understand common side effects of drugs and their statistical information, flexibly retrieve relevant medical reports, or annotate their own medical texts with our automated annotation tool. Supported by NLP technology and enriched with interactive visual components, we provide researchers and practitioners with a one-stop information analysis, retrieval, and annotation service. |
| Type Of Technology | Webtool/Application |
| Year Produced | 2024 |
| Impact | DrugWatch was published as a demo paper in ACL 2024, the top NLP conference. Paper link: https://aclanthology.org/2024.acl-demos.18/ |
| Title | NarrativePlay: Interactive Narrative Understanding |
| Description | NarrativePlay is a novel system that allows users to role-play a fictional character and interact with other characters in narratives in an immersive environment. We leverage Large Language Models (LLMs) to generate human-like responses, guided by personality traits extracted from narratives. The system incorporates auto-generated visual display of narrative settings, character portraits, and character speech, greatly enhancing the user experience. Our approach eschews predefined sandboxes, focusing instead on main storyline events from the perspective of a user-selected character. NarrativePlay has been evaluated on two types of narratives, detective and adventure stories, where users can either explore the world or increase affinity with other characters through conversations. |
| Type Of Technology | Webtool/Application |
| Year Produced | 2024 |
| Impact | NarrativePlay was published as a demo paper in EACL 2024, one of the leading conferences in NLP. |
| Title | PANACEA: An Automated Misinformation Detection System on COVID-19 |
| Description | Our web-based misinformation detection system PANACEA on COVID-19 related claims has two modules, fact-checking and rumour detection. The fact-checking module, which is supported by novel natural language inference methods with a self-attention network, outperforms state-of-the-art approaches. It is also able to give automated veracity assessment and ranked supporting evidence with the stance towards the claim to be checked. In addition, PANACEA adapts the bi-directional graph convolutional networks model, which is able to detect rumours based on comment networks of related tweets, instead of relying on the knowledge base. This rumour detection module assists by warning the users in the early stages when a knowledge base may not be available. |
| Type Of Technology | Webtool/Application |
| Year Produced | 2023 |
| Open Source License? | Yes |
| Impact | Our developed system has been summarised in a demo paper accepted in EACL 2023, a leading conference in NLP. It has also been selected to present in AI UK 2023. |
| Title | Topic-Driven and Knowledge-Aware Transformer for Dialogue Emotion Detection |
| Description | Transformer encoder-decoder for emotion detection in dialogues |
| Type Of Technology | Software |
| Year Produced | 2021 |
| Open Source License? | Yes |
| Impact | Since its recent release, the software available on GitHub has already been starred several times. |
| URL | https://zenodo.org/record/6352567 |
| Title | Topic-Driven and Knowledge-Aware Transformer for Dialogue Emotion Detection |
| Description | Transformer encoder-decoder for emotion detection in dialogues |
| Type Of Technology | Software |
| Year Produced | 2022 |
| Open Source License? | Yes |
| URL | https://zenodo.org/record/6352566 |
| Description | A webinar at the Gillmore Centre for Financial Technology, Warwick Business School |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | I was invited to give a talk on the recent developments in NLP for (1) automatic construction of hierarchical topic taxonomy from a large-scale text corpus; (2) named entity recognition and financial event extraction from text; (3) sentiment analysis including sentiment classification, aspect-based sentiment analysis and contrastive opinion extraction; and (4) end-to-end NLP approaches for trading signals detection. I concluded my talk with an outlook on potential NLP technologies that will shape the future of FinTech. The talk has sparked a few follow-up discussions on using NLP in the finance domain. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://www.wbs.ac.uk/events/view/7787 |
| Description | Featured in Futurum, an online magazine |
| Form Of Engagement Activity | A magazine, newsletter or online publication |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Schools |
| Results and Impact | Yulan He was featured in Futurum Careers, an online magazine, discussing her work on teaching computers to understand human language and offering guidance to young people interested in AI and NLP. Futurum Careers is a free online resource and magazine aimed at introducing 14-19-year-olds worldwide to the world of work in science, tech, engineering, maths, medicine, social sciences, humanities and the arts for people and the economy. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://futurumcareers.com/teaching-computers-to-understand-our-language |
| Description | Interview by New Scientist |
| Form Of Engagement Activity | A press release, press conference or response to a media enquiry/interview |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Media (as a channel to the public) |
| Results and Impact | I have been interviewed by New Scientist to comment on the ChatGPT detector. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://www.newscientist.com/article/2355035-chatgpt-detector-could-help-spot-cheaters-using-ai-to-w... |
| Description | Invited talk at AI UK 2022 |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Presented my work on machine reasoning for natural language understanding in AI UK 2022. My talk led to a collaborative project with AQA and joint research proposals with a few UK universities. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://www.turing.ac.uk/node/7396 |
| Description | Invited talk at Google |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Industry/Business |
| Results and Impact | Yulan He was invited to give a talk at Google on event-centric text understanding. Follow-on activities were planned for future collaboration. |
| Year(s) Of Engagement Activity | 2022 |
| Description | Invited talk at LSEG |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Invited talk on "Advancing FinTech through NLP Research" at the London Stock Exchange Group in January 2024. |
| Year(s) Of Engagement Activity | 2023 |
| Description | Invited talk at Zebra Technologies |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | Invited talk on "Interactive Narrative Understanding" at Zebra Technologies in November 2023. |
| Year(s) Of Engagement Activity | 2023 |
| Description | Invited talk at the Queen-Mary University of London |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Postgraduate students |
| Results and Impact | Yulan He was invited to give a talk on "Incorporating Commonsense Knowledge for Emotion Analysis" at the Queen-Mary University of London in June 2021.The talk sparked questions and discussion. |
| Year(s) Of Engagement Activity | 2021 |
| Description | Invited talk at the University of Cambridge |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Postgraduate students |
| Results and Impact | Yulan He was invited to give a talk on "Hierarchical Interpretation of Neural Text Classification" in the Language Technology Lab (LTL) at the University of Cambridge, headed by Anna Korhonen and Nigel Collier. A follow-up discussion was held in March between Anna and Yulan to explore potential future collaborations. |
| Year(s) Of Engagement Activity | 2022 |
| URL | http://131.111.150.181/talk/index/170564 |
| Description | Invited talk at the University of Durham |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Postgraduate students |
| Results and Impact | I present our recent work for (1) multimodal unreliable news detection using supervised contrastive learning; (2) COVID-related claim veracity assessment with a self-attention network built on natural language inference; and (3) vaccine attitude detection in social media through disentangled learning of stance and aspect topics. The talk sparked further discussions on potential collaborations. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://aihs.webspace.durham.ac.uk/seminars/ |
| Description | Keynote at CIKM 2023 |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Delivered a keynote presentation on "Interpretable Natural Language Understanding" at the 32nd ACM International Conference on Information and Knowledge Management (CIKM), which was held in Birmingham, UK in October 2023. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://uobevents.eventsair.com/cikm2023/yulan-he |
| Description | Keynote at INLG 2024 |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Postgraduate students |
| Results and Impact | A keynote on Enhancing LLM Reasoning through Reflection and Refinement was given at the 17th International Natural Language Generation Conference held in Tokyo, Japan in September 2024. |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://2024.inlgmeeting.org/keynotes.html |
| Description | Keynote at MATHMOD 2025 |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Postgraduate students |
| Results and Impact | Gave a plenary talk on "Advanced in Interpretable Language Modelling" at the 11th Vienna International Conference on Mathematical Modelling (MATHMOD 2025). |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://www.mathmod.at/ |
| Description | Keynote at NLDB 2023 |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Delivered a keynote on "Interpretable Language Understanding" at the 28th International Conference on Natural Language & Information Systems (NLDB), held in Derby, UK in June 2023. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://www.derby.ac.uk/events/latest-events/nldb-2023/ |
| Description | Panel member of Human-Like AI Forum, organised by the Cardiff Centre for AI, Robotics & Human-Machine Systems (IROHMS), University of Cardiff. |
| Form Of Engagement Activity | A formal working group, expert panel or dialogue |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | I was invited to participate a panel of Human-Like AI Forum, organised by the Cardiff Centre for AI, Robotics & Human-Machine Systems (IROHMS), University of Cardiff, in Feb. 2021. We have discussed how the the research in AI changed in the last 5 years, what is the biggest challenge in each of our respective AI field at the moment, and what will happen in the next 5 to 10 years in AI in general. |
| Year(s) Of Engagement Activity | 2021 |
| Description | Tutorial at DeepLearn 2024 Summer School and the Westlake University AI Open Course. |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Postgraduate students |
| Results and Impact | Gave a tutorial on Machine Reading Comprehension with Large Language Models in DeepLearn 2024 and the Westlake University AI Open Course. |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://deeplearn.irdta.eu/2024/ |
| Description | Tutorial in the Oxford Machine Learning Summer School |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Postgraduate students |
| Results and Impact | Yulan He was invited to give a tutorial on recent developments in sentiment analysis in the Oxford Machine Learning Summer School which was held in August 2021. The tutorial has attracted over 200 participants. As participants highly praised the tutorial, Yulan was invited to give a tutorial again in the Summer School in August 2022. |
| Year(s) Of Engagement Activity | 2021 |
| URL | https://www.oxfordml.school/2021 |
| Description | Tutorial in the Oxford Machine Learning Summer School |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Postgraduate students |
| Results and Impact | I delivered a tutorial on recent developments in sentiment analysis in the Oxford Machine Learning Summer School, targeting postgraduate students and researchers working in AI and machine learning. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://www.oxfordml.school/oxml2022 |
| Description | invited talk at Alan Turing Institute |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Postgraduate students |
| Results and Impact | Yulan He was invited to give a talk on topic-driven sentiment analysis in the NLP interest group at Alan Turing Institute. The talk has attracted some interests from the audience and potential future collaborations were planned. |
| Year(s) Of Engagement Activity | 2022 |