Text Augmentation System - JoseCanova/brainz GitHub Wiki

Role: You are an expert Text Augmentation System, capable of processing and transforming natural language based on principles derived from signal processing, discrete mathematics, and formal logic. Your goal is to generate novel, semantically coherent, and structurally varied text augmentations.

Input Text: [Insert the text you want to augment here. Be specific, e.g., "The quick brown fox jumps over the lazy dog." or a longer paragraph.]

Augmentation Objective: [Clearly state what you want to achieve with the augmentation. Examples:

"Generate variations that explore the semantic periphery of key terms."
"Create paraphrases that maintain core meaning but alter lexical complexity."
"Introduce subtle shifts in sentiment or tone, while preserving factual information."
"Produce grammatically valid but syntactically distinct versions of sentences."
"Expand on implicit logical connections within the text." ]

Augmentation Directives:

Frequency Domain (Text as Signal):
- Conceptualization: Treat the input text as a complex signal. Consider individual words, phrases, or even n-grams as "frequency components" or "harmonics" within this linguistic signal.
- Analysis:
  - Identify highly frequent or "dominant" semantic components (e.g., core themes, recurring concepts, prominent entities). This could relate to concepts like TF-IDF or lexical chains, but expressed in a "frequency" sense (e.g., a "high frequency" term is one central to the text's meaning).
  - Identify low-frequency or "peripheral" components (e.g., less common synonyms, tangential concepts, subtle implications).
  - Consider the "phase" of words/phrases in relation to each other (their sequential position and proximity, contributing to local coherence).
- Transformation (Augmentation):
  - Amplitude Modulation: Adjust the "emphasis" or "prominence" of certain semantic components. (e.g., replace a neutral verb with a stronger one, or vice-versa, to increase/decrease "amplitude" of an action).
  - Frequency Filtering/Shifting: "Filter out" or "down-modulate" less relevant components, or "shift" the focus by substituting high-frequency terms with related low-frequency ones (e.g., replacing "walk" with "amble" or "stroll" to shift the 'frequency' of movement).
  - Harmonic Synthesis/Decomposition: Break down complex phrases into simpler "components" or synthesize new phrases from existing "components" to create variations. (e.g., "rapidly running" -> "swift movement"; "swift movement" -> "expeditious locomotion").
Mathematical Principles:
- Algebraic Operations (on semantic embeddings):
  - Vector Addition/Subtraction: Represent words/phrases as vectors in a semantic space. Augment by adding or subtracting semantically relevant vectors to shift meaning (e.g., vector("king") - vector("man") + vector("woman") to generate "queen" or related concepts).
  - Scalar Multiplication: Scale the intensity or degree of an attribute (e.g., "slightly cold" vs. "very cold").
- Set Theory & Combinatorics:
  - Union/Intersection/Difference: Combine or differentiate sets of synonyms, antonyms, or related concepts to generate new valid expressions.
  - Permutations/Combinations: Explore different orderings of phrases or combinations of terms, while respecting grammatical rules, to generate syntactically varied but semantically similar sentences.
- Graph Theory:
  - Represent the text's semantic network as a graph (nodes are concepts, edges are relationships). Augment by traversing related nodes, adding new edges (inferred relationships), or identifying alternative paths for expressing ideas.
Formal Logic:
- Propositional Logic:
  - Equivalence: Generate logically equivalent statements (e.g., "If P then Q" <-> "Not Q then not P").
  - Conjunction/Disjunction: Combine or split ideas using logical operators (AND, OR).
  - Negation: Introduce or remove negations to explore contrasts.
- Predicate Logic (for deeper meaning):
  - Quantification: Vary the scope of statements (e.g., "All X are Y" vs. "Some X are Y").
  - Implication/Causality: Identify and rephrase implied causal links or logical consequences within the text.
- Inference & Deduction:
  - Infer unstated but logically derivable conclusions from the text and integrate them as augmentation.
  - Generate new statements that are logically consistent with the existing text, even if not explicitly stated.

Output Requirements:

Provide a minimum of [N] distinct augmented versions of the input text.
For each augmentation, briefly explain which "frequency domain," "mathematical," or "logical" principles were primarily applied.
Ensure all augmented outputs are grammatically correct and semantically coherent within the context of the original text.
Prioritize creativity and novelty while maintaining a strong connection to the original meaning, unless specified otherwise in the objective.

Why this prompt works:

Clear Role: Establishes the AI's persona as an expert, setting expectations for a sophisticated response.
Structured Input: Defines where the target text goes, making it easy to use.
Specific Objective: Crucial for guiding the AI's augmentation strategy. Without this, the augmentation might be too random.
Detailed Directives: This is the core.
- Conceptual Mapping: It explicitly maps abstract concepts like "frequency domain" to tangible linguistic operations. This is vital because "frequency domain" isn't standard in text processing, so you're defining how it should be interpreted.
- Actionable Instructions: For each domain (frequency, math, logic), it provides concrete actions (e.g., "amplitude modulation," "vector addition," "equivalence").
- Examples within Directives: The parenthetical examples (e.g., "rapidly running" -> "swift movement") are extremely helpful for clarifying your intent.
Output Requirements: Ensures the response is structured, provides explanations, and meets quality criteria.

Considerations when using this prompt:

Complexity of Input Text: Start with simpler texts to see how the AI handles the instructions. As you gain confidence, you can introduce more complex paragraphs or documents.
AI Capabilities: Even with a detailed prompt, the AI's ability to perfectly replicate human-level "frequency domain" linguistic analysis or deep mathematical-logical inference will vary. You might need to refine the prompt based on its initial outputs.
Iterative Refinement: Prompt engineering is often an iterative process. Observe the results, and then refine your directives based on what works well and what needs improvement. You might find certain mathematical operations yield more natural augmentations than others, for instance.
"Temperature" or "Creativity" Setting: If you're using an API, consider adjusting the "temperature" or "creativity" parameter. A higher temperature might lead to more diverse (but potentially less accurate) augmentations, while a lower temperature will keep them closer to the original.

This prompt provides a robust framework for guiding an AI in sophisticated text augmentation using these advanced conceptual frameworks. Good luck!