Movie Recommendation Assistant

Problem

Users often provide vague constraints such as "sci-fi", "Nolan", or "high-rated new films", while IMDB, Douban, and Baidu Baike can block direct scraping. Scores, reviews, and canonical links are also scattered across Chinese and English sources, so a single page scrape or a single-language search is fragile.

How It Works

• Intent parsing: extract title, genre, director, actors, and preferences; fall back to high-rated popular films when no keyword is present.
• Parallel search: run Chinese and English WebSearch requests together, then pull IMDB, Douban scores and tt_id / subject_id from snippets.
• Detail enrichment: prefer Rotten Tomatoes for Western titles, and use search snippets to fill Chinese scores and reviews.
• Translation and ranking: translate English synopses, reviews, and genre labels into Chinese; rank by keyword match and score when constraints exist, otherwise filter IMDB >= 7.0 or Douban >= 7.0.

Boundaries

Missing scores are shown as unavailable and missing reviews are omitted. If show_widget is unavailable, output degrades to plain text. The workflow is suitable for candidate recommendation, score aggregation, and Chinese review summaries, not for manual row-level validation across film databases.