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.
2
datasets available to search
ShareScore release 0.9.0
Dataset results
2 results for “Comparative sentences”
Dataset for training classifiers of comparative sentences
<p><br> As there was no large publicly available cross-domain dataset for comparative argument mining, we create one composed of sentences, potentially annotated with BETTER / WORSE markers (the first object is better / worse than the second object) or NONE (the sentence does not contain a comparison of the target objects). The BETTER sentences stand for a pro-argument in favor of the first compared object and WORSE-sentences represent a con-argument and favor the second object. </p> <p>We aim for minimizing dataset domain-specific biases in order to capture the nature of comparison and not the nature of the particular domains, thus decided to control the specificity of domains by the selection of comparison targets. We hypothesized and could confirm in preliminary experiments that comparison targets usually have a common hypernym (i.e., are instances of the same class), which we utilized for selection of the compared objects pairs. </p> <p>The most specific domain we choose, is computer science with comparison targets like programming languages, database products and technology standards such as Bluetooth or Ethernet. Many computer science concepts can be compared objectively (e.g., on transmission speed or suitability for certain applications). The objects for this domain were manually extracted from List of-articles at Wikipedia. In the annotation process, annotators were asked to only label sentences from this domain if they had some basic knowledge in computer science. </p> <p>The second, broader domain is brands. It contains objects of different types (e.g., cars, electronics, and food). As brands are present in everyday life, anyone should be able to label the majority of sentences containing well-known brands such as Coca-Cola or Mercedes. Again, targets for this domain were manually extracted from `List of''-articles at Wikipedia.</p> <p>The third domain is not restricted to any topic: random. For each of 24~randomly selected seed words 10 similar words were collected based on the distributional similarity API of JoBimText (http://www.jobimtext.org). Seed words created using randomlists.com: book, car, carpenter, cellphone, Christmas, coffee, cork, Florida, hamster, hiking, Hoover, Metallica, NBC, Netflix, ninja, pencil, salad, soccer, Starbucks, sword, Tolkien, wine, wood, XBox, Yale.</p> <p>Especially for brands and computer science, the resulting object lists were large (4493 in brands and 1339 in computer science). In a manual inspection, low-frequency and ambiguous objects were removed from all object lists (e.g., RAID (a hardware concept) and Unity (a game engine) are also regularly used nouns). The remaining objects were combined to pairs. For each object type (seed Wikipedia list page or the seed word), all possible combinations were created. These pairs were then used to find sentences containing both objects. The aforementioned approaches to selecting compared objects pairs tend minimize inclusion of the domain specific data, but do not solve the problem fully though. We keep open a question of extending dataset with diverse object pairs including abstract concepts for future work. </p> <p>As for the sentence mining, we used the publicly available index of dependency-parsed sentences from the Common Crawl corpus containing over 14 billion English sentences filtered for duplicates. This index was queried for sentences containing both objects of each pair. For 90% of the pairs, we also added comparative cue words (better, easier, faster, nicer, wiser, cooler, decent, safer, superior, solid, terrific, worse, harder, slower, poorly, uglier, poorer, lousy, nastier, inferior, mediocre) to the query in order to bias the selection towards comparisons but at the same time admit comparisons that do not contain any of the anticipated cues. This was necessary as a random sampling would have resulted in only a very tiny fraction of comparisons. Note that even sentences containing a cue word do not necessarily express a comparison between the desired targets (dog vs. cat: He's the best pet that you can get, better than a dog or cat.). It is thus especially crucial to enable a classifier to learn not to rely on the existence of clue words only (very likely in a random sample of sentences with very few comparisons). For our corpus, we keep pairs with at least 100 retrieved sentences.</p> <p>From all sentences of those pairs, 2500 for each category were randomly sampled as candidates for a crowdsourced annotation that we conducted on figure-eight.com in several small batches. Each sentence was annotated by at least five trusted workers. We ranked annotations by confidence, which is the figure-eight internal measure of combining annotator trust and voting, and discarded annotations with a confidence below 50%. Of all annotated items, 71% received unanimous votes and for over 85% at least 4 out of 5 workers agreed -- rendering the collection procedure aimed at ease of annotation successful.</p> <p>The final dataset contains 7199 sentences with 271 distinct object pairs. The majority of sentences (over 72%) are non-comparative despite biasing the selection with cue words; in 70% of the comparative sentences, the favored target is named first.</p> <p>You can browse though the data here: https://docs.google.com/spreadsheets/d/1U8i6EU9GUKmHdPnfwXEuBxi0h3aiRCLPRC-3c9ROiOE/edit?usp=sharing </p> <p>Full description of the dataset is available in the <a href="https://arxiv.org/abs/1809.06152">workshop paper at ACL 2019 conference</a>. Please cite this paper if you use the data: Franzek, Mirco, Alexander Panchenko, and Chris Biemann. "Categorization of Comparative Sentences for Argument Mining." <em>arXiv preprint arXiv:1809.06152</em> (2018).</p> <pre>@inproceedings{franzek2018categorization, title={Categorization of Comparative Sentences for Argument Mining}, author={Panchenko, Alexander and Bondarenko, and Franzek, Mirco and Hagen, Matthias and Biemann, Chris}, booktitle={Proceedings of the 6th Workshop on Argument Mining at ACL'2019}, year={2019}, address={Florence, Italy} }</pre> <p> </p> <p> </p>
Portuguese Comparative Sentences: A Collection of Labeled Sentences on Twitter and Buscapé
<p>More and more customers demand online reviews of products and comments on the Web to make decisions about buying a product over another. In this context, sentiment analysis techniques constitute the traditional way to summarize a user’s opinions that criticizes or highlights the positive aspects of a product. Sentiment analysis of reviews usually relies on extracting positive and negative aspects of products, neglecting comparative opinions. Such opinions do not directly express a positive or negative view but contrast aspects of products from different competitors. </p> <p>Here, we present the first effort to study comparative opinions in Portuguese, creating two new Portuguese datasets with comparative sentences marked by three humans. This repository consists of three important files: (1) lexicon that contains words frequently used to make a comparison in Portuguese; (2) Twitter dataset with labeled comparative sentences; and (3) Buscapé dataset with labeled comparative sentences.</p> <p>The lexicon is a set of 176 words frequently used to express a comparative opinion in the Portuguese language. In these contexts, the lexicon is aggregated in a filter and used to build two sets of data with comparative sentences from two important contexts: (1) Social Network Online; and (2) Product reviews.</p> <p>For Twitter, we collected all Portuguese tweets published in Brazil on 2018/01/10 and filtered all tweets that contained at least one keyword present in the lexicon, obtaining 130,459 tweets. Our work is based on the sentence level. Thus, all sentences were extracted and a sample with 2,053 sentences was created, which was labeled for three human manuals, reaching an 83.2% agreement with Fleiss' Kappa coefficient. For Buscapé, a Brazilian website (https://www.buscape.com.br/) used to compare product prices on the web, the same methodology was conducted by creating a set of 2,754 labeled sentences, obtained from comments made in 2013. This dataset was labeled by three humans, reaching an agreement of 83.46% with the Fleiss Kappa coefficient.</p> <p>The Twitter dataset has 2,053 labeled sentences, of which 918 are comparative. The Buscapé dataset has 2,754 labeled sentences, of which 1,282 are comparative.</p> <p><strong>The datasets contain these labeled properties:</strong></p> <ul> <li> <p><em>text</em>: the sentence extracted from the review comment.</p> </li> <li> <p><em>entity_s1: </em>the first entity compared in the sentence<em>.</em></p> </li> <li> <p><em>entity_s2: </em>the second entity compared in the sentence.</p> </li> <li> <p><em>keyword: </em>the comparative keyword used in the sentence to express comparison.</p> </li> <li> <p><em>preferred_entity: </em>the preferred entity.</p> </li> <li> <p><em>id_start: </em>the keyword's initial position in the sentence.</p> </li> <li> <p><em>id_end: </em>the keyword's final position in the sentence.</p> </li> <li> <p><em>type: </em>the sentence label, which specifies whether the phrase is a comparison<em>.</em></p> </li> </ul> <p><strong>Additional Information:</strong></p> <p><em><strong>1 </strong>- </em>The sentences were separated using a sentence tokenizer.</p> <p><em><strong>2 </strong>- </em>If the compared entity is not specified, the field will receive a value: "__".</p> <p><em><strong>3</strong> - </em>The property <em>"type"</em> can contain five values, they are:</p> <ul> <li> <p><em>0: Non-comparative (</em>Não Comparativa<em>)</em>.</p> </li> <li> <p><em>1: Non-Equal-Gradable (</em>Gradativa com Predileção<em>)</em>.</p> </li> <li> <p><em>2: Equative</em> <em>(</em>Equitativa<em>).</em></p> </li> <li> <p><em>3: Superlative (</em>Superlativa<em>).</em></p> </li> <li> <p><em>4: Non-Equal-Gradable (</em>Não Gradativa<em>).</em></p> </li> </ul> <p> </p> <p>If you use this data, please cite our paper as follows: </p> <p><em>"Daniel Kansaon, Michele A. Brandão, Julio C. S. Reis, Matheus Barbosa,Breno Matos, and Fabrício Benevenuto. 2020. Mining Portuguese Comparative Sentences in Online Reviews. In Brazilian Symposium on Multimedia and the Web (WebMedia ’20), November 30-December 4, 2020, São Luís, Brazil. ACM, New York, NY, USA, 8 pages. https://doi.org/10.1145/3428658.3431081"</em></p> <p>--------------</p> <p><strong>Plus Information:</strong></p> <p>We make the raw sentences available in the dataset to allow future work to test different pre-processing steps. Then, if you want to obtain the exact sentences used in the paper above, you must reproduce the pre-processing step described in the paper (<em>Figure 2</em>). </p> <p>For each sentence with more than one keyword in the dataset: </p> <ul> <li>You need to extract three words before and three words after the comparative keyword, creating a new sentence that will receive the existing value in the “<em>type</em>” field as a label;</li> <li>The original sentence will be divided into <em>n</em> new sentences. (<em>n</em>) is the number of keywords in the sentence;</li> <li>The stopwords should not be accounted for as part of this range (<em>3 words</em>);</li> </ul> <p>Note that: the final processed sentence can have more than six words because the stopwords are not counted as part of the range.</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.