Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
116
datasets available to search
ShareScore release 0.9.0
Dataset results
116 results for “Natural Language”
The Alice Dataset: fMRI Dataset to Study Natural Language Comprehension in the Brain
Open the record for dataset details and reuse information.
Study Data: Is It Time to Reconsider our Current Approaches to Natural Language Understanding?
<p>Participants consisted of 95 traditional, undergraduate students enrolled in multiple undergraduate psychology courses offered at a private, Mid-Atlantic liberal arts college.</p>
Natural Language-Guided Programming User Study
<p>In this dataset you find the user study data that was used in the <strong><em>Natural Language-Guided Programming</em></strong> paper, which is accepted for Onward! 2021. A preprint can be found here <a href="https://arxiv.org/pdf/2108.05198.pdf">https://arxiv.org/pdf/2108.05198.pdf</a>. The dataset consists of the following files:</p> <ul> <li> <p>benchmark.json contains 201 test cases. Each test case consists of context, a natural language intent and target code. The test cases are intended to evaluate a model that can predict code giving a piece of context code and a natural language intent. The test cases were derived from Jupyter notebooks that were crawled from Github projects with permissive licenses. In the project_metadata field you find information about the original project such as its git url and license.</p> </li> <li> <p>predictions-annotated.json contains predictions of the three models used in the paper for 100 test cases in benchmark.json. Each prediction is accompanied with qualitive assesments from three annotators.</p> </li> <li> <p>train-index.jsonl is the list of github projects that were used for training the models.</p> </li> <li> <p>eval-index.jsonl is a list of github projects that we kept separate for evaluation. The benchmark.json was created from a random subset of the projects in this list.</p> </li> </ul> <p>For more details we refer to the paper.</p>
lilGym: Natural Language Visual Reasoning with Reinforcement Learning, model files
<p>Baselines models for the paper <a href="https://lil.nlp.cornell.edu/lilgym"><em>lil</em>Gym: Natural Language Visual Reasoning with Reinforcement Learning</a>.</p>
Report on Transformers interpretability for Natural Language Processing: A case study on Technical Debt classification
<p>Transformer models have significantly advanced the field of natural language processing (NLP), achieving exceptional results in various tasks. However, these models are often seen as "black boxes", providing limited insight into the factors influencing their predictions. It has become crucial to develop and utilise methods for interpreting and explaining these models to uncover their complex inner workings. This report discusses the latest techniques and tools that aid in a more profound understanding of transformer models within NLP. Additionally, it explores a vital industrial use case: Technical Debt (TD) classification. In this context, the report leverages transformer model interpretability tools and Retrieval Augmented Generation (RAG) to analyse and understand the characteristics of text in Github issues, distinguishing between TD and non-TD.</p> <p>This report thoroughly outlines an approach to improve the transparency and reproducibility of machine learning models, with a special emphasis on TD classification. It integrates the RAG approach and exploits feature attribution techniques, presenting a route to create AI systems that are not only high-performing but also demonstrably trustworthy and comprehensible. Through a detailed examination of word patterns in TD classification and the innovative use of the RAG approach, the research highlights a strong dedication to promoting transparency and responsibility in AI systems, potentially ushering in a new phase in machine learning research that focuses on clarity and dependability.</p>
Variational Inference for Learning Representations of Natural Language Edits
<p>Performance Evaluation of Edit Representations (PEER), the dataset we use in the paper <a href="https://arxiv.org/abs/2004.09143">"Variational Inference for Learning Representations of Natural Language Edits"</a>.</p>
Files and code for English dictionaries, gold and silver standard corpora for biomedical natural language processing related to SARS-CoV-2 and COVID-19
<p><span lang="EN-GB">Automated information extraction with natural language processing (NLP) tools is required to gain systematic insights from the large number of COVID-19 publications, reports and social media posts, which far exceed human processing capabilities. </span></p> <p><span lang="EN-GB">Here we present an NLP toolbox comprising COVID-19-related dictionaries and annotated corpora in English as well as useful code and workflows for their update and use. The dictionaries contain terms referring to the COVID-19 disease, the SARS-CoV-2 virus, its variants and common mutations, respectively. They were used together with the EasyNER NLP tool to extract and annotate all 764 398 abstracts in the CORD-19 dataset, creating a very large silver standard corpus (named Lund-Annotated-CORD-19 corpus). This was complemented with a small gold standard corpus consisting of PubMed abstracts manually annotated for key entity classes such as disease, virus, symptom, protein/gene, cell type, chemical and species terms. </span></p> <p><span lang="EN-GB">The toolbox can support various text analysis tasks related to COVID-19 such as named entity recognition and co-mention analysis. A preliminary version of the toolbox, which was released early in the pandemic, was</span><span lang="EN-GB"> for example already used to create a COVID-19 knowledge graph and study the evolution and variation of COVID-19-related terminology. In addition, the toolbox can be applied in the development of other NLP tools, for example to train and evaluate large language models.</span></p> <p><span lang="EN-GB">When using the toolbox, please cite this record and the associated article.</span></p> <p> </p> <p> </p>
Custom language model checkpoints used in "Testing the limits of natural language models for predicting human language judgments"
<p>Checkpoint files for an RNN, LSTM, BILSTM and n-gram models used the paper "Testing the limits of natural language models for predicting human language judgments"</p>
Technical Debt Classification in Issue Trackers using Natural Language Processing based on Transformers
<p>In order to ensure transparency and reproducibility, we have made everything available publicly here, including the Code, Models, Datasets and more. All the files and their functionality used in this paper are explained clearly in the <strong>README.md</strong> file.</p> <p>Background: Technical Debt (TD) needs to be controlled and tracked during software development. Support to automatically track TD in issue trackers is limited. </p> <p>Aim: We explore the usage of a large dataset of developer-labeled TD issues in combination with cutting-edge Natural Language Processing (NLP) approaches to automatically classify TD in issue trackers.</p> <p>Method: We mine and analyze more than 160GB of textual data from GitHub projects, collecting over 55,600 TD issues and consolidating them into a large dataset (GTD dataset). We use such datasets to train and test Transformer ML models. Then we test the model's generalization ability by testing them on six unseen projects. Finally, we re-train the models including part of the TD issues from the target project to test their adaptability. </p> <p>Results and Conclusion: (i) We create and release the GTD dataset, a comprehensive dataset including TD issues from 6,401 public repositories with various contexts; (ii) By training Transformers using the GTD dataset, we achieve performance metrics that are promising; (iii) Our results are a significant step forward towards supporting the automatic classification of TD in issue trackers, especially when the models are adapted to the context of unseen projects after fine-tuning.</p>
Figure 11. Comparison of MultiNet and MST representations-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>In the sentence (7) there are two clauses describing two hypothetical situations. These two<br> situations are connected by the relation COND in MultiNet representation (Figure 11).</p>
Figure 10. Comparison of MultiNet and MST representations.-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>Helbig [18, p. 507] defines the expression (s MCONT c) as “a specification of the<br> informational or mental content c of a mental or informational process s… By default, the second<br> argument c is assumed to be a hypothetical object or situation.”<br> MCONT relation properties are roughly equivalent to those of subject-verb complexes<br> (SVSBs) of MST. MCONT relation can generally be treated as capturing the idea behind what<br> philosophers call propositional attitudes in representational theories of mind [12] or opaque<br> contexts [10] in semantics. For example, consider the sentence in (3) rewritten below as (6) and<br> semantic representation of which is given in Figure 10.</p>
Figure 8. Comparison of MultiNet and MST representations-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>Figure 8 shows the representation of the<br> first reading of the sentence (5) in MWR. The MultiNet is shown at top of the figure, and the corresponding mental space representation is shown at the bottom side. Blue circles mark<br> hypothetical or nonreal objects and situations and red circles mark real objects and situations.<br> Facticity value for John is real, for unicorn it is nonreal, and for the process of riding, marked with<br> blue broken circle, it is non-real as well.</p>
Figure 7. The mental spaces set up by the sentence If John buys the car, he will drive to Berlin.-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>In the sentence (4) there are three proper names which constitute elements of the base space<br> or reality space (R). The CNSB if sets up the hypothetical space (H) with elements identical to those<br> of reality space (Figure 7). Every space’s internal structure is presented in the boxes next to them.<br> In the next part, we will see how MultiNet’s built-in meaning representation mechanisms are<br> capable of representing the basic principles of mental space building outlined above.</p>
Figure 6. The mental spaces set up by the sentence Mary thinks that John smokes.-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>The proper nouns Mary and John setup a base space (B). By the help of background<br> knowledge and activated frames we know that they are names of female and male humans. Not<br> having access to the previous discourse, we also consider their existence presupposed. The SVSB<br> Mary thinks that sets up a belief space (L) relative to space B (Figure 6). The identity connector<br> maintains the referential link between elements a and a ׳ both referring to the same person.</p>
Figure 9. Comparison of MultiNet and MST representations-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>Second interpretation of the sentence (5) can be represented with changing John’s Facticity<br> attribute-value from [FACT = real] to [FACT = nonreal] making the whole situation and its<br> elements non-real (Figure 9).</p>
Figure 5. The mental spaces set up by the second interpretation of the sentence in the film, John is riding a unicorn.-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>Second interpretation: The proper name John exists only in film space without having a<br> counterpart in base space. Therefore, the second interpretation of the sentence is: John is a<br> film character who is riding a unicorn in the film (Figure 5).</p>
Figure 4. The mental spaces set up by the first interpretation of the sentence in the film, John is riding a unicorn-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>First interpretation: The reality space (let’s call it R) contains an element a associated with<br> the proper name John. The noun phrase a unicorn introduces an element b ׳ to the film space<br> (call it F). I is the connector linking a in the space B to a ׳ in the space F (Figure 4). Since the<br> elements of both mental spaces are co-referential, this connector is an identity connector.<br> The rectangles represent the internal structure of the spaces next to them. The dashed line<br> indicates that the space F is set up in relation to R and that it is subordinate to R in<br> discourse.</p>
Figure 1. The upper ontology of sorts in MultiNet (after Helbig [17])-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>One of the distinguishing features of MultiNet is its commitment to the Cognitive Adequacy<br> requirement. According this requirement [9], semantic representations and knowledge<br> representations should be centered around concepts. Concepts2 are represented by nodes in the<br> graphical representation of the network. Every node belongs to a specific sort defined by the<br> MultiNet’s ontology of sorts (Figure 2).</p>
Figure 1. Mental space representation of "In the play, Mary is excited"-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>Thus an entity can have a variable reality status depending on the mental<br> space to which it belongs. The mental space constructed by the sentence Poirot is a Belgian<br> detective is a non-real imaginary story space of which Poirot is an element. But when we say in<br> reality, Poirot is not Belgian the constructed space is reality space in which Poirot (the actor, not<br> the character) does not have the fictional nationality.</p>
Figure 2. Semantic representation of the sentence Peter finished the discussion in MultiNet after Helbig [8, p. 447].-Representing Mental Spaces and Dynamics of Natural Language Semantics
<p>In Figure 3, semantic frame of the concept Finish realized in the form of the verb finish<br> requires two C-roles: An agent represented by the relation AGT, and an affected entity represented<br> by the relation AFF. Here agent is Peter and the affected entity is an abstract object ([SORT = ad]<br> means the concept is a dynamic abstraction).</p>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.