Mastering Drupal AI: Extracting Structured Data from Articles with AI Automators скачать в хорошем качестве

Mastering Drupal AI: Extracting Structured Data from Articles with AI Automators

Learn how to turn messy, free-text article bodies into clean, queryable Drupal fields—author, email, publication month, and year—using AI Automators. Justin walks through spinning up the Drupal CMS AI demo on DrupalForge, creating content types and fields, configuring taxonomy vocabularies, and building a reliable chain of automators (text extraction → entity reference → HTML cleanup). You’ll see real-world debugging, model/provider choices, batch vs. direct execution, and a glimpse of scaling this approach to PDFs with Unstructured, plus ideas for views (author pages, issue pages) and vector search improvements. Chapters 00:00 Problem: unstructured articles & goals for structured fields 00:31 Why structure matters: SEO, Views, and better vector search 02:28 Author pages & issue-level browsing vision 03:19 Plan: use AI Automators + Author content type (entity reference) 06:00 Spin up Drupal CMS AI demo on DrupalForge (keys, VS Code, setup) 07:48 Environment prep: enable dev tools, Composer/Drush updates 10:10 Create Article content type & core fields via the chatbot 13:52 Two-step extraction: intermediary text fields → entity reference 16:52 Handling LLM hiccups: retries, clearing history, model choices 21:01 Create Month taxonomy; ordering months (not alphabetical) 24:25 Publication fields: month (taxonomy) & year (taxonomy) 27:01 Build Author content type (bio, email, site, photo) 33:57 Author entity reference field + “create referenced entities” 36:11 Enable AI modules: Automators, Agents, Explorer, Logging, etc. 37:18 Automators: extract author name (weights & run order) 40:33 Automator for email extraction 41:31 Automators for month & four-digit year 43:21 Advanced automator: map extracted names ↔ emails to Author nodes 46:26 Batch vs Direct; cron/Q settings for production stability 47:25 Picking models/providers (LLM proxy, 40 vs 40-mini, etc.) 49:54 Cleanup automator: refactor HTML, remove author/email/issue from body 52:22 Review automator run order (weights & dependencies) 55:19 Live test: run, debug AJAX errors, fix configuration 1:05:46 Fix bundle mismatch (Article vs Author) and re-test 1:09:44 Success: fields filled, Author node created, body cleaned 1:10:24 Views: author landing pages & issue groupings 1:13:13 Avoiding duplicate authors (code tweak/bug note) 1:18:27 Teaser: PDFs → HTML with Unstructured + automators 1:20:18 Logging AI requests: when to enable and how to manage 1:22:37 Wrap-up & next steps Credits: Salim (Sal) Lakhani,   / sklakhani   Justin Keiser,   / keiserjb   Infrastructure, tooling, and AI provider by https://devpanel.com/ #ai #RAG #VectorSearch #Views #DevEnvironments #VSCode #Composer #AIIntegration #AIChatbot #AIInnovation #FreelyGive #DevPanel #DrupalForge