v1.2.3

Changelist

• Keyword Ordering in Evaluation View: Added the ability to order keywords in the evaluation view by default order, alphabetically, or by the value of any metric. This enhancement helps identify the best and worst performing keywords.

  

v1.2.2

Changelist

• Editable Grades in Any Evaluation Status: You can now edit any grade in an evaluation at any status, even for already completed evaluations. All corresponding metrics will be automatically recalculated. This feature allows for post-analysis adjustments to correct any mistakenly assigned grades or questionable assessments.

• Bug Fixes and Minor Interface Improvements: Resolved several issues to improve application stability and performance, and made minor user interface improvements for a more intuitive experience.

v1.2.1

Changelist

• Compact Metrics: Added a compact mode for displaying metric values. Clicking on a metric toggles between compact and full display modes. The selected mode is saved for the current browser.

  

v1.2.0

Changelist

• New Scale Type "Detail" with Corresponding Metrics: Introduced a new scale type called Detail, which allows grading search results on a scale from 1 to 10. Corresponding metrics for this scale include:

  • CG(d)@k - Cumulative Gain
  • DCG(d)@k - Discounted Cumulative Gain
  • nDCG(d)@k - Normalized Discounted Cumulative Gain
  

• Support for Multiple Scale Types in Evaluations: Evaluations are no longer restricted to metrics of the same scale type (binary, graded, or detail). You can now include metrics of different types within a single evaluation. If an evaluation includes metrics from different scales, you need to define Transformers, a new feature that allows converting grades from one scale to another. The scale and transformers are set during the creation of the evaluation and cannot be changed after it starts.

  

• Various Bug Fixes and Minor Improvements: Resolved several issues to improve application stability and performance, and made minor user interface improvements for a more intuitive experience.

v1.1.0

Changelist

• Metric Changes with Baseline Comparison: Metrics can now be viewed as percentage changes relative to previous metrics, providing clearer insights into performance trends. Users can set a "Baseline" evaluation for metric comparisons, allowing for consistent benchmarking. By default, metrics are compared with the previous finished evaluation of the same metrics type.

  

• Filter Evaluation Keywords by Query: Added the ability to search within evaluations using keywords. This feature makes it easier to locate specific keyword search results and metrics.

  

• Archive Evaluations: Introduced the option to archive evaluations. Archived evaluations are removed from the main evaluations list and are excluded from the model metrics progress chart. A filtering panel allows users to display all evaluations, only active ones, or only archived ones, enhancing navigation and organization.

  

• "Pin to Top" Evaluations: Added the option to "Pin to Top" any evaluation. Pinned evaluations will appear at the beginning of the evaluations list for quick and easy access.

  

• Evaluation Timestamps: Evaluations now display their creation and finish timestamps, providing better tracking and historical reference.

  

• Various Bug Fixes and Minor Improvements: Resolved several issues to improve application stability and performance, and made minor user interface improvements for a more intuitive experience.

v1.0.2

Changelist

• Delete Pending Evaluation: Users now have the option to delete any pending evaluations, providing more flexibility and control.

• Updated Tag Matching Logic: The tag matching logic has been updated from OR to AND. This means users will now only be able to give feedback to search evaluations if they have all the required tags assigned to them.

• Previous Feedback Review: Users can now click the Previous button to review and re-evaluate previously graded search snapshot. This feature allows users to easily correct any mistakenly given feedback, improving the overall user experience.

  

v1.0.1

Changelist

• Keyboard Shortcuts for Evaluator Buttons: Users can now interact with the evaluator's interface more efficiently by using keyboard shortcuts. Pressing number keys (0-9) will trigger the corresponding evaluation buttons, allowing for faster and more streamlined workflows.

  

• Dynamic Visual Feedback on Button Press: Introduced dynamic visual feedback when evaluation buttons are pressed. This enhancement improves user experience by providing immediate visual confirmation of button interactions, making the interface more intuitive and responsive.

v1.0.0

SearchTweak v1.0.0 - First Open Source Release

We are excited to announce the first open-source release of SearchTweak, a comprehensive web application designed to evaluate and optimize search engine results. This initial release includes the core functionality of SearchTweak, providing powerful tools for:

• Search Results Optimization: Analyze and improve the relevance and performance of search engine outcomes.
• Data Labeling: Use SearchTweak for creating labeled datasets for machine learning models.
• Advanced Analytics: Generate insights with built-in analytics tools to make data-driven decisions.
• Customization and Extensibility: SearchTweak is built on Laravel, allowing developers to easily extend or customize the functionality to suit their needs.
• REST API: Makes it easy to integrate SearchTweak into your data pipeline and connect it with your existing tools and services.

Key Features:

• Search relevance evaluation tools.
• Labeling functionality for training machine learning models.
• Easy-to-use web interface for advanced analytics.
• Built on Laravel for easy customization and flexibility.
• REST API for seamless integration into your existing workflows.
• Docker support for seamless setup and deployment.

This release marks the beginning of SearchTweak as an open-source project, empowering developers and businesses to improve their search capabilities and adapt the platform to their unique requirements.

We welcome feedback and contributions! Feel free to open issues, submit pull requests, and help shape the future of SearchTweak.