Promptology. Season 2. Part 7.
Prompts Structure.
This is one of the main posts of promptology. Without a structure in place, it is impossible to put together the published pieces of prompts into a coherent whole (see season 1 and 2 of promptology).
To understand:
- It took more than 2 years
- Four people participated in the test. Positions thousands of experiments.
- Spent about $20000.
- Dozens of buckets of coffee have been drunk.
Block 0: Presets and Global Rules
Sets the fundamental constraints and quality standards for all work.
0.1 Content Focus: Sets the main purpose of the content
0.2 SEO Goals: Defines the initial list of keys
0.3 Content Quality sub-block (including STOP WORDS): Hard rules that eliminate «water,» clichés, value judgments, and ambiguity. This is the core for creating clean, factual text.
0.4. Language Settings: Specifies the language and dialect
0.5. Humanization sub-block: A set of rules for simulating natural, «human» speech. Called for specific sections.
0.6. Creativity sub-block: Creative content generation algorithm. Called for specific sections.
Step 1: Preparing the information
Task: Analyze and additionally collect raw data.
- Analyzing and Collecting Entities: Defining TSS. Extending the semantic field through related, implicit and comparative entities.
- Analyzing Intents: Identifying and categorizing user search intentions.
- Building the Knowledge Graph and NER: Creating a semantic graph to identify relationships and content gaps. Using Named Entity Recognition to classify entities into functional categories.
- Data Validation.
Step 2: Building a taxonomy
Task: Design a logical and SEO-efficient content structure based on Stage 1 data.
- Structure development: Creating a hierarchy of sections (H2, H3) focused on intents. Application of different principles (e.g. Q&A, optimization for Featured Snippets, etc.).
- Advanced Structural Blocks: For example introducing «Semantic Triangulation», «View from the Other Side» and so on.
- Integration of Creative Blocks
- Validation and Inference: Checking that the taxonomy coverage of entities and intents is at least 90%.
Step 3: Create new content
Task: Filling the designed structure with text.
- A set of instructions of 30-50 points on text creation (from the subject): salience of entities, E-E-A-T, no «water», uniqueness, focus on details, and so on.
- Linking and Structuring: Rules for integrating links into content. Creating comparative sections and tables.
- Deep Analytics Blocks: Integration of unique elements (from the topic), for example, «Expert Advice», «Quantification Principle» (backing up theses with figures) and so on.
- Final Check and Audit: Semantic Proximity Check. Audit of linking with PASS/FAIL system.
Step 4: Generate a summary table
Task: Create a concise, structured squeeze of key data (on strictly defined parameters, cross-checking the validity of the data).
Step 5: Generating a short description
Task: Creating a capacious and attractive snippet (short description).
Step 6: Creating FAQs and Micro Markups
1. Creating FAQs according to given rules and cross-checking data.
2. HTML generation: Create HTML code for FAQ with correct markup Schema.org.
3. Generating JSON-LD: Creating micro-markup to improve search engine submission.
Step 7: Create Meta Tags and Content for Images
Task: Finalizing content for publication.
- Meta tags: Create SEO-optimized Title and Description.
- Image Content: Provide guidelines for embedding images, including file names, alt-texts, and unique, interesting captions.
Data Block: [DATABASE].
Challenge: Directory to provide relevant internal re-linking.
This is a universal structure. It is sharpened for each task separately.