AI Film Recommender DLE

AI Film Recommender for DataLife Engine

AI Film Recommender adds a dedicated page to your site where visitors find films and TV series by describing what they want in their own words. A request can describe a mood, a genre, an atmosphere, a plot, a country of production, a release period or any combination of these.

The module does not import third-party content and never shows titles that are not in your site’s database. The AI builds a list of recommendations, then the module looks for the matching DLE posts and shows only the ones it finds.

How it works

1. The visitor sends a text request through the module’s AJAX interface.
2. The request is combined with the site’s catalogue settings and the administrator’s instructions.
3. The selected AI provider returns a structured list: original title, localised titles and release year.
4. The module matches this data with DLE posts using the configured additional fields and the post title.
5. The posts it finds are rendered through the standard show.short.php mechanism and the selected TPL template.

Supported AI providers

• OpenAI (GPT) — its own API key, model selection, web search, forced web search and adjustable search context depth.
• Google Gemini — its own API key, model selection and web search through Google Search grounding.
• DeepSeek — its own API key and model selection; built-in web search is not used for this provider.

The list of allowed models is stored inside the module. If an unknown or outdated model is selected, the provider’s default model is used instead.

Prompt and catalogue theme

The administrator sets the context in which the AI builds its recommendations. Ready-made profiles are available:

• films and TV series from different countries;
• Korean dramas;
• anime;
• Turkish series;
• Indian cinema;
• a custom profile.

Any profile can have its own catalogue description and extra instructions. This lets the recommendations reflect the site’s niche, the language of the titles, what is in the database and your own rules.

Matching recommendations with the DLE database

• Ordered title fields. The administrator picks several additional fields and sets their priority. The original title field goes first, followed by localised titles and the post title as a fallback.
• Release year check. Pick one or more year fields and choose strict, preferred or disabled matching.
• Structured AI response. The search takes the original title, alternative titles and the year into account — not just a single line of text.
• Deterministic choice. When there are several matches, the priority of the title source, the year match and the post ID decide.
• XFields compatibility. The xfields.txt format is supported for older DLE versions, xfields.json and DLEXFields for newer ones.

Displaying results

• Results are built from real posts on your site.
• Cards can use a separate TPL template without touching the main shortstory.tpl.
• Standard DLE tags and additional fields processed through show.short.php are supported.
• The number of recommendations shown is set in the admin panel.
• The module page has its own SEO Title, Description and Keywords.

Caching and request limits

• AI responses are stored in a structured JSON cache keyed by provider, model, web search mode, site profile and other request parameters.
• Different profiles and settings never share a cache, even if the visitor sent exactly the same text.
• The cache of earlier versions is read only in a compatible context and migrated to the new format.
• Each user group gets its own hourly request limit.
• The counter increases only after a result has been received and built successfully.
• The admin panel shows the cache status and lets you clear the module’s cache safely.

Proxies and network requests

A shared proxy can be enabled for requests to AI APIs. HTTP, SOCKS5 and SOCKS5H are supported, with username and password authentication. SOCKS5H resolves domain names through the proxy.

URLs and SEO routes

• The module is always available at the internal address index.php?do=ai-film-recommender.
• You can enable a separate SEO-friendly prefix for the page, for example /ai-films/.
• On DLE 19.0 and newer, the module’s rules are synchronised with rules.json.
• During synchronisation the module checks for conflicts, makes a backup and changes only the rules that belong to it.
• For older DLE versions the panel shows the rules to add to your web server configuration by hand.

Admin panel

The control panel is built from standard DLE components and has three sections:

• Settings — providers, models, web search, proxy, site profile, field matching, results template, limits and SEO;
• Cache management — cache file information, checks and cleanup;
• URLs and SEO routes — setting up and synchronising SEO-friendly URLs.

API keys and the proxy password are not displayed after saving. Admin actions check access rights and the user_hash token.

Compatibility and requirements

• DataLife Engine: 14.x–20.0.
• PHP: 7.4–8.4.
• PHP extensions: cURL, JSON, mbstring.
• AI API: a key for at least one supported provider.
• Content database: for accurate matching, at least one real additional field with the original title must be configured; a year field is recommended but optional.

Good to know

• The quality of results depends on what is in your database, how completely titles and years are filled in, the selected model and the visitor’s request.
• Web search is used only by providers that support it.
• The module shows only recommendations that match posts on your site.
• The editable configuration file is created when you save the settings and is not part of the installation archive, so updating the module never overwrites your site’s settings.

Video preview