1// Code node: Clean and enhance query
2function preprocessQuery(query) {
3  // Clean
4  let clean = query.trim();
5  
6  // Expand abbreviations
7  const expansions = {
8    "API": "Application Programming Interface",
9    "RAG": "Retrieval Augmented Generation",
10    "LLM": "Large Language Model"
11  };
12  
13  // Add context for better retrieval
14  return {
15    originalQuery: query,
16    cleanQuery: clean,
17    searchQuery: clean // Can add synonyms or rephrase
18  };
19}
20
21return { json: preprocessQuery($json.query) };

1// OpenAI node: Enhance query for better retrieval
2const enhancePrompt = `
3Rephrase this query to improve search results. 
4Generate 3 alternative phrasings that capture the same intent.
5
6Original query: "${$json.query}"
7
8Return JSON:
9{
10  "queries": ["original", "rephrasing1", "rephrasing2"]
11}`;

1// Vector Store Search node
2// Top K: 10 (retrieve more, filter later)
3// Score Threshold: 0.6
4
5// With multi-query: search with each rephrased query
6// then deduplicate results

1// Code node: Deduplicate results from multi-query
2const allResults = $input.all();
3const seen = new Set();
4const unique = [];
5
6for (const result of allResults) {
7  const content = result.json.document.pageContent;
8  const hash = content.substring(0, 100); // Simple dedup
9  
10  if (!seen.has(hash)) {
11    seen.add(hash);
12    unique.push(result);
13  }
14}
15
16// Sort by score descending
17unique.sort((a, b) => (b.json.score || 0) - (a.json.score || 0));
18
19// Keep top 5
20return unique.slice(0, 5);

1// OpenAI node: Re-rank results by relevance
2const rerankPrompt = `
3Given the question and search results, rank the results by relevance.
4
5Question: "${$json.query}"
6
7Results:
8${$json.results.map((r, i) => `[${i}] ${r.content}`).join('\n\n')}
9
10Return JSON array of indices ordered by relevance: [most_relevant_index, ...]
11Only include indices of chunks that actually help answer the question.`;

1// Code node: Build optimized context
2const query = $json.query;
3const topResults = $json.rankedResults.slice(0, 5);
4
5const context = topResults.map((r, i) => {
6  const source = r.metadata.source || "unknown";
7  return `--- Source ${i + 1}: ${source} ---
8${r.content}`;
9}).join('\n\n');
10
11const systemPrompt = `You are a helpful assistant that answers questions 
12based on the provided context. 
13
14Rules:
151. Only use information from the context below
162. If the answer is not in the context, say "I could not find this information in the knowledge base"
173. Cite sources when possible: [Source 1], [Source 2]
184. Be concise but complete
195. Use Vietnamese for the response`;
20
21const userPrompt = `Context:
22${context}
23
24Question: ${query}
25
26Answer:`;
27
28return { json: { systemPrompt, userPrompt, sourceCount: topResults.length } };

1// OpenAI Chat node
2// System Message: {{ $json.systemPrompt }}
3// User Message: {{ $json.userPrompt }}
4// Model: gpt-4o-mini
5// Temperature: 0.3 (lower for factual answers)
6// Max Tokens: 500

1// Code node: Format response
2const answer = $json.message.content;
3const sources = $json.sources;
4
5const response = {
6  answer: answer,
7  sources: sources.map(s => ({
8    title: s.metadata.source,
9    relevance: s.score
10  })),
11  confidence: sources[0]?.score || 0,
12  timestamp: new Date().toISOString()
13};
14
15return { json: response };

1// Code node: Check if results are relevant
2const results = $json.searchResults;
3const threshold = 0.7;
4
5const relevant = results.filter(r => r.score >= threshold);
6
7if (relevant.length === 0) {
8  return {
9    json: {
10      answer: "I could not find relevant information for your question.",
11      fallback: true,
12      suggestion: "Try rephrasing or ask a more specific question."
13    }
14  };
15}
16
17return { json: { results: relevant, fallback: false } };