Find research datasets worth reusing

opencc-by-4.0Mar 2007View details →

ANE Site Placemarks for Google Earth

ANE.kmz is a set of site placemarks for Google Earth of a selection of the most important archaeological sites in the Ancient Near East. ANE.kmz works with Google Earth Pro, which first has to be downloaded for free. When opened inside Google Earth Pro, ANE.kmz gives, to the left, an alphabetic list of ancient sites and, to the right, on the satellite images the same sites marked. For the moment, there are some 2500 sites with modern names; among them some 400 have ancient names. Additions of more sites are planned. Ancient names are written without parenthesis. Modern names are within parenthesis. Most sites have been identified on the satellite images. ANE Waters.kmz is an experimental set of provisional water placemarks for Google Earth covering Mesopotamia up to modern time. ANE Picture.jpg is just illustrating the appearence of ANE.kmz before zooming in and is not for use.

opencc-by-4.0Sep 2022View details →

Extended datasets from MM-IMDB and Ads-Parallelity dataset with the features from Google Cloud Vision API

This is extended datasets from MM-IMDB [<a href="https://openreview.net/forum?id=S12_nquOe">Arevalo+ ICLRW'17</a>], Ads-Parallelity [<a href="https://arxiv.org/abs/1807.08205">Zhang+ BMVC'18</a>] dataset with the features from Google Cloud Vision API. These datasets are stored in jsonl (JSON Lines) format. Abstract (from our paper): There is increasing interest in the use of multimodal data in various web applications, such as digital advertising and e-commerce. Typical methods for extracting important information from multimodal data rely on a mid-fusion architecture that combines the feature representations from multiple encoders. However, as the number of modalities increases, several potential problems with the mid-fusion model structure arise, such as an increase in the dimensionality of the concatenated multimodal features and missing modalities. To address these problems, we propose a new concept that considers multimodal inputs as a set of sequences, namely, deep multimodal sequence sets (DM2S2). Our set-aware concept consists of three components that capture the relationships among multiple modalities: (a) a BERT-based encoder to handle the inter- and intra-order of elements in the sequences, (b) intra-modality residual attention (IntraMRA) to capture the importance of the elements in a modality, and (c) inter-modality residual attention (InterMRA) to enhance the importance of elements with modality-level granularity further. Our concept exhibits performance that is comparable to or better than the previous set-aware models. Furthermore, we demonstrate that the visualization of the learned InterMRA and IntraMRA weights can provide an interpretation of the prediction results. Dataset (MM-IMDB and Ads-Parallelity): We extended two multimodal datasets, namely, MM-IMDB [<a href="https://openreview.net/forum?id=S12_nquOe">Arevalo+ ICLRW'17</a>], Ads-Parallelity [<a href="https://arxiv.org/abs/1807.08205">Zhang+ BMVC'18</a>] for the empirical experiments. The MM-IMDB dataset contains 25,925 movies with multiple labels (genres). We used the original split provided in the dataset and reported the F1 scores (micro, macro, and samples) of the test set. The Ads-Parallelity dataset contains 670 images and slogans from persuasive advertisements to understand the implicit relationship (parallel and non-parallel) between these two modalities. A binary classification task is used to predict whether the text and image in the same ad convey the same message. We transformed the following multimodal information (i.e., visual, textual, and categorical data) into textual tokens and fed these into our proposed model. We used the <a href="https://cloud.google.com/vision">Google Cloud Vision API</a> for the visual features to obtain the following four pieces of information as tokens: (1) text from the OCR, (2) category labels from the label detection, (3) object tags from the object detection, and (4) the number of faces from the facial detection. We input the labels and object detection results as a sequence in order of confidence, as obtained from the API. We describe the visual, textual, and categorical features of each dataset below. MM-IMDB: We used the title and plot of movies as the textual features, and the aforementioned API results based on poster images as visual features. Ads-Parallelity: We used the same API-based visual features as in MM-IMDB. Furthermore, we used textual and categorical features consisting of textual inputs of transcriptions and messages, and categorical inputs of natural and text concrete images.

opencc-by-4.0Sep 2022View details →

Google's Audioset: Reformatted

<pre>Google's AudioSet consistently reformatted During my work with Google's AudioSet(https://research.google.com/audioset/index.html) I encountered some problems due to the fact that Weak (https://research.google.com/audioset/download.html) and Strong (https://research.google.com/audioset/download_strong.html) versions of the dataset used different csv formatting for the data, and that also labels used in the two datasets are different (https://github.com/audioset/ontology/issues/9) and also presented in files with different formatting. This dataset reformatting aims to unify the formats of the datasets so that it is possible to analyse them in the same pipelines, and also make the dataset files compatible with psds_eval, dcase_util and sed_eval Python packages used in Audio Processing. For better formatted documentation and source code of reformatting refer to https://github.com/bakhtos/GoogleAudioSetReformatted -Changes in dataset All files are converted to tab-separated `*.tsv` files (i.e. `csv` files with `\t` as a separator). All files have a header as the first line. -New fields and filenames Fields are renamed according to the following table, to be compatible with psds_eval: Old field -> New field YTID -> filename segment_id -> filename start_seconds -> onset start_time_seconds -> onset end_seconds -> offset end_time_seconds -> offset positive_labels -> event_label label -> event_label present -> present For class label files, `id` is now the name for the for `mid` label (e.g. `/m/09xor`) and `label` for the human-readable label (e.g. `Speech`). Index of label indicated for Weak dataset labels (`index` field in `class_labels_indices.csv`) is not used. Files are renamed according to the following table to ensure consisted naming of the form `audioset_[weak|strong]_[train|eval]_[balanced|unbalanced|posneg]*.tsv`: Old name -> New name balanced_train_segments.csv -> audioset_weak_train_balanced.tsv unbalanced_train_segments.csv -> audioset_weak_train_unbalanced.tsv eval_segments.csv -> audioset_weak_eval.tsv audioset_train_strong.tsv -> audioset_strong_train.tsv audioset_eval_strong.tsv -> audioset_strong_eval.tsv audioset_eval_strong_framed_posneg.tsv -> audioset_strong_eval_posneg.tsv class_labels_indices.csv -> class_labels.tsv (merged with mid_to_display_name.tsv) mid_to_display_name.tsv -> class_labels.tsv (merged with class_labels_indices.csv) -Strong dataset changes Only changes to the Strong dataset are renaming of fields and reordering of columns, so that both Weak and Strong version have `filename` and `event_label` as first two columns. -Weak dataset changes -- Labels are given one per line, instead of comma-separated and quoted list -- To make sure that `filename` format is the same as in Strong version, the following format change is made: The value of the `start_seconds` field is converted to milliseconds and appended to the `filename` with an underscore. Since all files in the dataset are assumed to be 10 seconds long, this unifies the format of `filename` with the Strong version and makes `end_seconds` also redundant. -Class labels changes Class labels from both datasets are merged into one file and given in alphabetical order of `id`s. Since same `id`s are present in both datasets, but sometimes with different human-readable labels, labels from Strong dataset overwrite those from Weak. It is possible to regenerate `class_labels.tsv` while giving priority to the Weak version of labels by calling `convert_labels(False)` from convert.py in the GitHub repository. -License Google's AudioSet was published in two stages - first the Weakly labelled data (Gemmeke, Jort F., et al. "Audio set: An ontology and human-labeled dataset for audio events." 2017 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE, 2017.), then the strongly labelled data (Hershey, Shawn, et al. "The benefit of temporally-strong labels in audio event classification." ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2021.) Both the original dataset and this reworked version are licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) </pre> Class labels come from the AudioSet Ontology, which is licensed under CC BY-SA 4.0.

opencc-by-4.0Jul 2018View details →

Dateset: Capturing the influence of geopolitical ties from Wikipedia with reduced Google matrix

This dataset provides complementary material to the scientific research presented in the paper "Capturing the influence of geopolitical ties from Wikipedia with reduced Google matrix", accepted for publication in PLOS ONE under number PONE-D-18-07662R1. A draft version of the paper is available at <a href="https://arxiv.org/abs/1803.05336">https://arxiv.org/abs/1803.05336</a> This paper presents two studies targeting two groups of countries: <ul> <li>[40] the set 40 worldwide countries set ;</li> <li>[EU] the set of 27 European Union countries as of February 2013.</li> </ul> Data is derived using Reduced Google matrix analysis on the Wikipedia English, Arabic, Russian, German and French editions collected in February 2013. Networks representing each edition are available here: <a href="http://www.quantware.ups-tlse.fr/QWLIB/topwikipeople/index.html">http://www.quantware.ups-tlse.fr/QWLIB/topwikipeople/index.html</a> The following files are given: <ul> <li>[GRedured_40.zip] GReduced matrix and its decomposition for [40] countries set</li> <li>[GRedured_EU.zip] GReduced matrix and its decomposition for [EU] countries set</li> <li>[PageRank_vs_CheiRank.zip] PageRank versus CheiRank figures for RuWiki and ArWiki</li> <li>[Sensitivity.xlsx] and [Sensitivity_html.xlsx] Sensitivity values for both [EU] and [RU] in either .xlsx or in .html format</li> </ul>

opencc-by-4.0Jul 2023View details →

Papers on Google Scholar using "sonification, auditory display, audification, sonify" as search terms

Data set from a Google Scholar search in January 2023 on the terms "sonification, auditory display, audification, sonify" and added abstracts from various online ressources and keywords (automatically extracted from the abstracts only), containing: <ul> <li>their title,</li> <li>a website/URL (as referenced by Google scholar),</li> <li>author(s),</li> <li>publisher information,</li> <li>their google rank in our search,</li> <li>publication year,</li> <li>the number of citations;</li> <li>paper abstracts;</li> <li>keywords generated from abstracts.</li> </ul>

edi48/100

Supervised land cover classification using Google Earth Engine in Córdoba, Argentina, 2018-2020

Land cover information is critical to scientific, economic, and public policy-making. There is a high demand for accurate and timely land cover information that affects the accuracy of all subsequent applications. The availability of Google Earth Engine (GEE), which derives temporal aggregation methods from time-series images (i.e., the use of metrics such as mean or median), has also enabled optimization of computation time, such as managing large amounts of data to obtain more accurate results. Our objective was to obtain a land cover map for the northwest of the province of Córdoba, Argentina. The study was carried out in rural communities that belong to the departments of Cruz del Eje and Ischilín, northwest of Córdoba, and have different degrees of intervention in the land cover. Sentinel 2 Level 2A images were acquired for the study area. Images available from January 1, 2018, to December 31, 2020, were sampled. To create a thematic map, the median value was calculated for the sample of images from the selected time interval. Finally, the Normalized Difference Vegetation Index (NDVI) was calculated and added to the total bands of the median image. Training polygons were placed there considering the visual features in the median image. The Random Forest algorithm was used as the classification method. To verify the quality of the classified map, a list of 97,753 verification pixels was obtained. In addition, a confusion matrix was created to collect the conflicts that arise between categories, and the precision and kappa coefficient was calculated to define the quality of the map obtained. Image acquisition, preprocessing, and analysis were performed on the Google Earth Engine platform. Thematic maps with eight classes were obtained, with a total area of 719880 ha. The confusion matrix showed an overall precision of 99.26% and a corrected kappa index of 0.99, the classes were correctly classified by the algorithm.

openCC (other)Dec 2023View details →

opencc-by-4.0Nov 2020View details →

African Swine Fever Worldwide Epidemiology Data - OIE Webscrape example - Geocoded using Google API and Manual

Example African Swine Fever dataset generated by programs described in following publication  Title: Web-scraping programmatic techniques in aggregating difficult to access OIE WAHIS animal disease outbreak information; using African Swine Fever in Europe as an example. Short running title: Methods for web-scraping OIE WAHIS data. Abstract: This study describes and makes available new methods for acquiring difficult to access, publicly available, disease surveillance data. It uses World Organisation for Animal Heath (OIE) data on African Swine Fever (ASF) outbreaks in Belarus and its neighbouring European countries to showcase the importance of adequate disease surveillance data to inform decision-making. The data acquired from these methods allow for large-scale, geospatial outbreak mapping and summary statistics of any terrestrial disease listed on the OIE World Animal Health Information System (WAHIS) database. These techniques will make important epidemiological data more accessible to the scientific community and aid in gaining further insight into the occurrence and spread of OIE listed diseases in a timely manner, fulfilling an important function of disease surveillance.

opencc-zeroJun 2015View details →

Google Trends and Wikipedia Page Views

Abstract (our paper) The frequency of a web search keyword generally reflects the degree of public interest in a particular subject matter. Search logs are therefore useful resources for trend analysis. However, access to search logs is typically restricted to search engine providers. In this paper, we investigate whether search frequency can be estimated from a different resource such as Wikipedia page views of open data. We found frequently searched keywords to have remarkably high correlations with Wikipedia page views. This suggests that Wikipedia page views can be an effective tool for determining popular global web search trends. Data personal-name.txt.gz: The first column is the Wikipedia article id, the second column is the search keyword, the third column is the Wikipedia article title, and the fourth column is the total of page views from 2008 to 2014. personal-name_data_google-trends.txt.gz, personal-name_data_wikipedia.txt.gz: The first column is the period to be collected, the second column is the source (Google or Wikipedia), the third column is the Wikipedia article id, the fourth column is the search keyword, the fifth column is the date, and the sixth column is the value of search trend or page view. Publication This data set was created for our study. If you make use of this data set, please cite: Mitsuo Yoshida, Yuki Arase, Takaaki Tsunoda, Mikio Yamamoto. Wikipedia Page View Reflects Web Search Trend. Proceedings of the 2015 ACM Web Science Conference (WebSci '15). no.65, pp.1-2, 2015. http://dx.doi.org/10.1145/2786451.2786495 http://arxiv.org/abs/1509.02218 (author-created version) Note The raw data of Wikipedia page views is available in the following page. http://dumps.wikimedia.org/other/pagecounts-raw/

opencc-by-4.0Apr 2024View details →

Quién busca a quién algunas razones para no usar Google

La investigación se basa en un universo de análisis conformado por los 10 primeros resultados de búsqueda, para 10 términos de consulta, ejecutados por 30 personas en sesiones sucias en Google, y 1 persona en 2 sesiones controladas y limpias con diferentes características entre sí. El universo de análisis totaliza 3200 resultados de búsqueda, 3000 de sesiones sucias y 200 de sesiones limpias.  Cada uno de los resultados de búsqueda remiten a una URL/fuente. La unidad de análisis es la URL/fuente a la que remiten los resultados. Se clasifican por término de consulta, por posición/rango en la página de 10 resultados, y por caso (sesiones sucias sin control (Caso 1) o sesiones limpias controladas (Casos 2 y 3).  Las fuentes externas de datos son 30 participantes que aceptan realizar las tareas de búsqueda en Google y son seleccionados de acuerdo con su cumplimiento con los siguientes criterios: uso habitual de celular con tecnología Android y cuenta de Gmail activada en el dispositivo; Chrome como navegador por defecto; Google como buscador por defecto y Gmail como servicio de correo electrónico con el cual proveer identidad para poder acceder a descargas de cualquier AP.  Todas las sesiones sucias usan entonces el mismo navegador (Chrome, por defecto en los celulares), se ejecutan desde plataforma celular, y en el buscador Google geolocalizado para Argentina. Las sesiones limpias las ejecuta la investigadora en Google con un IP simulado de Perú a través de una red virtual, y en DuckDuckGo. Definición de casos de búsqueda: Ø  CASO 1: Búsqueda ejecutada por personas reales (30) con las siguientes características comunes: -Búsqueda en Google, plataforma celular, navegador Chrome, con cuenta de Google abierta (Sesión sucia: cuenta de Gmail, datos recopilados de historial de navegación e historial de búsquedas y sesión abierta, geolocalización activada). <a name="_Hlk126859255"></a>-Las 30 personas están geolocalizadas en La Plata, Buenos Aires, Argentina -Los mismos 10 términos de búsqueda. Ø  CASOS 2 y 3: Búsqueda ejecutada por 1 sola persona (la investigadora), con las siguientes características distintivas, definidas como caso 2 y caso 3: Ø  CASO 2: Búsqueda en Google, plataforma celular, navegador Chrome, sin cuenta de Google, desde un IP de Perú (Sesión limpia: sin cuenta de Gmail abierta, datos recopilados de historial de navegación e historial de búsquedas y sesión abierta eliminados, geolocalización activada). -Los mismos 10 términos de búsqueda. Ø  CASO 3: Búsqueda en DuckDuckGo, plataforma celular, navegador Chrome, sin cuenta de Google (Sesión limpia: sin cuenta de Gmail abierta, datos recopilados de historial de navegación e historial de búsquedas y sesión abierta eliminados, geolocalización activada). -Los mismos 10 términos de búsqueda. Términos de consulta Ø  2 términos de consulta controvertidos para los cuales hay dos posturas: matrimonio homosexual y pena de muerte, siguiendo a Gezici y otros, a Lemuth, a Leferink y a Krafft y otros, que buscan probar si ciertos temas controvertidos tienen más probabilidades de generar un resultado de búsqueda más seleccionado que temas más normales y no controvertidos partiendo de la hipótesis (en Leferink) de que la personalización tiene más influencia en las consultas controvertidas que en las no controvertidas. Ø  2 términos de consulta de eventos recientes: guerra Ucrania Rusia; persecuciones en Venezuela, siguiendo a Jiang, son términos de consulta para los cuales hay mucha publicación. Ø  2 términos de consulta de cola larga: tratamiento con hongos psilocybe, Operación lanza de Neptuno, siguiendo a Carreras Lario, a Jiang y a Neil, son términos de consulta que se han buscado poco. Ø  2 términos de consulta asociados a los servicios que brinda la empresa del buscador: correo electrónico y correo electrónico sin rastreadores, siguiendo a Edelman y Lockwood, son términos en los que el sesgo de contenido propio es factible de medir. Ø  2 términos comerciales de alta competencia empresarial: destinos turísticos; marcas de zapatillas deportivas, siguiendo a Carreras Lario y a Jiang.

opencc-by-4.0Apr 2022View details →

Google Speech Commands-Musan test set

This noisy speech test set is created from the Google Speech Commands v2 [1] and the Musan dataset[2]. It is introduced in our ICASSP 2022 paper [3].  Specifically, we created this test set by mixing the speech in the Google Speech Commands v2 test set with random noise in the Musan dataset at different signal to noise ratio -12.5,-10,0,10,20,30 and 40 decibel (dB).  The Google Speech Commands v2 dataset is under the Creative Commons BY 4.0 license. It could be downloaded at: http://download.tensorflow.org/data/speech_commands_v0.02.tar.gz The Musan dataset is under Attribution 4.0 International (CC BY 4.0). It could be downlowned at https://www.openslr.org/17/ Citations: [1] Pete Warden, “Speech commands: A dataset for limited-vocabulary speech recognition,” arXiv preprint arXiv:1804.03209, 2018. [2] David Snyder, Guoguo Chen, and Daniel Povey, “Musan: A music, speech, and noise corpus,” arXiv preprint arXiv:1510.08484, 2015. [3] V. A. Trinh, H. Salami Kavaki and M. I. Mandel, "Importantaug: A Data Augmentation Agent for Speech," ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022, pp. 8592-8596, doi: 10.1109/ICASSP43922.2022.9747003.

opencc-zeroJul 2022View details →

Web requests analysis of Italy websites which use Google Analytics

List of 504,038 domains of Italy found to contain Google Analytics. The front page for Italy-related domain names has been accessed through HTTPS or HTTP and analysed with webbkoll and jq to gather data about third-party requests, cookies and other privacy-invasive features. Together with the actual URL visited, the user/property ID is provided for 495,663 domains (extracted either from the cookies deposited or the URL of requests to Google Analytics). MX and TXT records for the domains are also provided. The most common ID found was 23LNSPS7Q6, with over 35k domains calling it (seemingly associated with italiaonline.it). The most common responding IP addresses were 3 AWS IPv4 addresses (over 40k domains) and 2 CloudFlare IPv6 addresses (over 12k domains).

opencc-by-4.0Sep 2022View details →

How do Google News' top 100 sources visually represent the data centres' energy footprint?

By querying "data centres' energy footprint" on Google News in incognito mode, the candidate has selected and mapped the top 100 results according to the ranking on May 15, 2022.

opencc-by-4.0May 2019View details →

Google Play Store Data

Mobile App Stores such as Google, Apple have wide range of applications to suffice every need of customers in the digital platform.  Customer feedback and ratings has always been one of the major metrics that can be used to review the performance and accordingly provide suitable recommendations to enhance the functionality. The Given dataset contain the feedback of the customer regarding the app used in app store. Data Set Column Details are as given below: Column name:  Description: Column Name in Working Sheet Datatype Please read the Readme.docs file

Mobile Cloud Computing Bibliographic Results from Google Scholar

This dataset contains all the results for the term "Mobile Cloud Computing" on Google Scholar until June 2018.  The data was acquired using Publish or Perish. The data has been cleaned such that the wrong and invalid results have been removed, duplicates have been removed. Titles are accurate and fine but authors and publishers info. etc. is still unclean. For textual analysis based on paper titles, this dataset is fine. For any other factor, such as institutional or journal or authorship analysis, this isn't a good choice.

Mobile Edge Computing Bibliographic Results from Google Scholar

This dataset contains all the results for the term "Mobile Edge Computing" on Google Scholar until June 2018.  The data was acquired using Publish or Perish. The data has been cleaned such that the wrong and invalid results have been removed, duplicates have been removed. Titles are accurate and fine but authors and publishers info. etc. is still unclean. For textual analysis based on paper titles, this dataset is fine. For any other factor, such as institutional or journal or authorship analysis, this isn't a good choice.

Edge Computing Bibliographic Results from Google Scholar

This dataset contains all the results for the term "Edge Computing" on Google Scholar until June 2018.  The data was acquired using Publish or Perish. The data has been cleaned such that the wrong and invalid results have been removed, duplicates have been removed. Titles are accurate and fine but authors and publishers info. etc. is still unclean. For textual analysis based on paper titles, this dataset is fine. For any other factor, such as institutional or journal or authorship analysis, this isn't a good choice.

Fog Computing Bibliographic Results from Google Scholar

This dataset contains all the results for the term "Fog Computing" on Google Scholar until June 2018.  The data was acquired using Publish or Perish. The data has been cleaned such that the wrong and invalid results have been removed, duplicates have been removed. Titles are accurate and fine but authors and publishers info. etc. is still unclean. For textual analysis based on paper titles, this dataset is fine. For any other factor, such as institutional or journal or authorship analysis, this isn't a good choice.

opencc-by-4.0May 2019View details →

Datos Autores USTA Google Scholar

Consolidado de los datos de las publicaciones de autores con filiación USTA (Colombia) en la plataforma Google Scholar. Parte del servicio de vigilancia tecnológica del Observatorio de Cienciometría de la Universidad Santo Tomás. Estrategia de búsqueda: “universidad santo tomas” OR “santo tomás university” OR “Univ Santo Tomas” OR “Universidad Santo Tomás” Herramientas:   Harzing’s Publish or Peris (POP), Microsoft Excel

opencc-zeroNov 2024View details →

Machine Learning Articles Extracted from Google Scholar

This dataset was created as part of a web scraping practice aimed at capturing academic information from Google Scholar. It contains data on Machine Learning research articles, including the article's title, authors, summary, direct link, citation count, and APA reference. This dataset was collected using Python and Selenium to develop skills in web scraping tools for extracting data from websites with dynamic content. The dataset was generated specifically as part of an academic exercise to learn and apply web scraping techniques, without a deep analysis intent for the data obtained. This dataset is intended as a resource for learning and evaluating the methods used in web data collection. Included Fields: <ul> <li>title: Title of the research article.</li> <li>link: Direct link to the article.</li> <li>authors: Names of the article’s authors.</li> <li>description: Summary or brief description of the article.</li> <li>citations: Number of times the article has been cited on Google Scholar.</li> <li>APA_citation: APA-formatted citation of the article.</li> </ul> This dataset was created solely for educational purposes and to demonstrate the application of web scraping techniques in a controlled environment.