Skip to content

CHANGELOG.md

Latest commit

 

History

History
63 lines (47 loc) · 2.37 KB

CHANGELOG.md

File metadata and controls

63 lines (47 loc) · 2.37 KB

Changelog

1.0.0 - 2026-03-21

First official Sentify release.

Overview

Sentify is a Flask web application for turning stock-related news into an actionable Buy, Hold, or Sell recommendation. It combines Yahoo Finance news retrieval, transformer-based sentiment analysis, article-level weighting, and an interactive dashboard into a single workflow.

Core capabilities

  • Search by company name or ticker symbol.
  • Autocomplete suggestions for company names and ticker symbols.
  • Fetch and review recent Yahoo Finance news for the selected company.
  • Score article sentiment and summarize the result into a Buy, Hold, or Sell recommendation with a confidence score.
  • Display per-article publish timing, sentiment output, and article details in the dashboard.

Recommendation controls

  • Adjustable news lookback window from the main dashboard.
  • Recommendation tuning controls for recency weighting.
  • Recommendation tuning controls for article content-length weighting.
  • Client-side recalculation support so weighting changes can update the recommendation view without reprocessing cached sentiment unnecessarily.

Dashboard and user experience

  • TradingView chart embed alongside the recommendation summary.
  • Search suggestion dropdown and improved company selection flow.
  • Clear cache action from the dashboard.
  • Recommendation summary cards and confidence display in the main interface.

Caching and fallback behavior

  • Cached Yahoo news list responses.
  • Cached article body retrieval.
  • Cached sentiment results keyed to the active model identity.
  • Stale-cache fallback when live Yahoo requests fail.
  • Cached-sentiment fallback when live model loading is unavailable.
  • Offline mode for local-only operation with stale cache reuse.
  • Cache retention and cleanup on startup.

Configuration and deployment behavior

  • Environment-variable based configuration for model selection, cache paths, cache TTLs, and weighting controls.
  • Local .env support.
  • Debug mode disabled by default for normal app runs.

Quality and reliability

  • Structured runtime logging in key fallback and failure paths.
  • Runtime hardening for cache cleanup and Yahoo response handling.
  • Automated unit test coverage for recommendation logic, fallback behavior, search behavior, and cache handling.
  • GitHub Actions CI coverage for both lint/pre-commit checks and the unittest suite.