OpenKnowledgeMaps Headstart Versions Save

Version 7 represents major progress in terms of future-proofing Head Start. It contains the result of a refactoring of the Head Start frontend in React and the introduction of a state-store based on Redux. The backend (data clients and machine learning pipeline) has been refactored into a microservice-architecture based on Docker containers, a Redis-queue and a Postgres database. In this release, knowledge map quality has been significantly improved by adding a default minimum-length for abstracts in BASE, improving the query sanitization and search behavior of data clients, as well as a range of metadata schema alignments, improved backward compatibility of the frontend, and various smaller bug fixes. The release also contains a new data client for Google Spreadsheets, which is the basis for CoVis. For more information on CoVis, please visit our projects page. Head Start has also been adapted to integrate with the new search-flow module resulting in a better search experience.

Major changes include:

• Refactoring of all visualization components in React #458, #456, #462, #481, #499, #501, #505, #509, #512, #513, #518, #525, #532, #535, #551, #551, #569, #570
• Containerization of R-based backend with Docker #427, #433, #503, #516, #517, #515, #517, #552, #590
• Addition of Google-Spreadsheets data client #428, #432, #461
• Integration with new search-flow module #539, #576
• Introduction of expressive error messages and search hints

Minor changes include:

• Introduction of test framework for Head Start client #444, #454
• Introduction of regression tests in backend #466
• Boolean queries for BASE #412
• Abstract length filter for BASE #464
• Data client retries on timeout #489
• AJAX based fallback behavior for search request timeouts #419
• Activation of browser back/forward buttons for zoom in/out #605
• Critical security and other maintenance updates

Bug Fixes:

• Re-enabled article types in PubMed #420, #479
• OpenAIRE data client bug fixes #450, #591
• Metadata schema bug fixes #440, #446, #494, #498, #514, #571, #579
• Streamgraph bug fixes #402
• Clustering bug fix #448
• Summarization stopwords bug fix #477
• Highlighting bug fix #496
• Visualization metadata backwards compatibility #530
• Document type updates for BASE #559
• Search query format bug fix #529
• Many other small bug fixes

Version 6 incorporates features developed in the collaboration project LinkedCat+. For more information on LinkedCat+, please visit our projects page. In addition, v6 includes minor changes and bug fixes that result in an improved user experience.

Major changes include:

• Implementation of streamgraphs
• Implementation of a browse-view prototype
• Workflow separation for knowledge maps and streamgraphs. Please note that with this change, data is no longer interchangeable between those #376

Minor changes include:

• Automatic merging of bubbles with an identical summary, regardless of cluster assignment #377
• Improved testing workflow to automatically test map data in a larger scale #368

Bugfixes:

• Fixed a bug where bubbles with one paper occasionally had no summary, #369
• The addition of “based” to the stopwords list improves summaries
• Fixed a bug where titles with invalid regex cause the map creation to fail, #346
• Critical security and other maintenance updates, #390

Version 5 incorporates features developed within two large collaboration projects: CRIS Vis and LinkedCat+. For more information on these projects, please visit our website. In addition, v5 includes many bugfixes and improvements that result in a better user experience.

Major changes include:

• Implementation of language-aware maps
• Enabling author-based maps
• A complete redesign of list entries, including the capability to add multiple sub-entries and images
• Enabling adding tags to bubble titles and list items, and an optional legend for these tags
• Consistent rotation of maps
• Better results for maps with only a few items
• Improved cluster naming

Version 5 also improves development, deployment, and maintenance of Headstart:

• A completely reworked event tracking system
• Split up of backend services
• Improved logging on backend
• Snapshot testing
• More specific errors when map creation fails
• Improved skin integration

The Visual Project Explorer (VIPER) is a unique open science application created by Open Knowledge Maps that provides visual overviews of research projects indexed by OpenAIRE. It enables funders, institutions and researchers to systematically explore a project’s output, and to understand its impact in different areas. VIPER exploits a unique property of OpenAIRE data: the link between projects and publications and datasets.

To realize VIPER, we have adapted and extended Head Start. New features include:

• A metrics component to enrich publication metadata with metrics from different sources. This is accompanied by a scaling functionality that rescales the map according to a given metric to enable a multi-faceted view on project outputs.
• An automatization component to automatically update project maps in certain intervals or when new resources have been added.
• An embed functionality to enable the inclusion of VIPER on external websites and in dashboards.

VIPER also comes with a dedicated search interface for selecting projects for visualization as an example.

Version 4 is a milestone release focusing on usability, stability, design, and usefulness of Headstart.

Major improvements include:

• Higher visibility for and easier access of open access items
• More context information on the map and visual hints for user actions
• Significant improvements in cluster naming
• Automatic creation of a snapshot of a map
• Improvements for mobile and tablet use
• Deduplication of entries
• Enable creating maps for the most recent documents for a search term
• Completion on more searches: better handling of edge cases and results with sparse information

As part of our mission of becoming ever more open, Headstart is now licensed under MIT 🎉

Version 3.5 introduces various new features and improvements both on the frontend and the backend, including

• Connection to the BASE Search Engine
• Vastly improved cluster naming procedures, esp. when no keywords are provided
• Possibility to add different skins for the visualization
• Reduction of force-directed layout processing time
• Lots of small improvements to the user experience on the frontend
• Heuristics for locating the PDF document on a landing page for display in Head Start

This release also includes a rewrite of the frontend with the following benefits:

• Improved modularization and decoupling in the frontend code base
• Simplification of visualizaton setup routines
• Improved debugging and deployment

Version 3 is our biggest release yet and represents a milestone for the project. It comprises of 347 individual commits.

Headstart 3 includes a wide variety of new and improved features:

• Integration of two new data sources: PubMed and DOAJ
• Responsive visualization
• Completely overhauled design
• Enhanced user experience
• Integration of Hypothes.is in the PDF preview

Headstart 3 also represents a major step in terms of front-end design, development and deployment, including:

• NPM for dependency management
• Webpack for packaging & deployment
• SASS and Handlebars to reduce the complexity of frontend code and to decouple appearance and functionality

Version 2.9 is an interim release on the way to Headstart 3. It represents a first look at many of the improvements that we have scheduled for version 3. A major architectural upgrade is that the visualization components have been decoupled with the help of the mediator pattern. We have also implemented many improvements to the visualization interface, such as search highlighting and support for multiple languages. The PLOS search has been extentded to include a number of options, and more descriptive labels for research areas using a TF-IDF labelling algorithm.

This version will be used in the initial release of Open Knowledge Maps.

Version 2.7 integrates a persistence and versioning system based on SQLite. Pre-processing can now be handled by passing JSON strings instead of using CSV files. See search_plos.php for a prototypical implementation of this behaviour - it queries the PLOS API for the 100 most relevant results for a search term and creates a visualization on the fly.

For a demo of this functionality, please go to http://openknowledgemaps.org/mozfest

Head Start is now part of the Open Knowledge Labs!

Version 2.6 mainly facilitates the integration of Head Start into other systems. A number of options have been introduced that make it easier to customize the visualization's behavior and appearance. By default, the screen real estate is now better utilized. In addition, several display issues have been fixed that occurred when the visualization was embedded in a existing website design. Finally, the system for academic evaluations introduced in v2.5 has been improved.

An adaptation of Head Start has been integrated in the conference scheduling system Conference Navigator 3 [registration required]. This version enables users to schedule papers directly from the visualization. Scheduled papers and recommended papers are highlighted.