mcp-tenant-credit-scorer/src/industry-matcher.js

MCP server for tenant credit scoring based on S&P corporate methodology

import { readFile } from 'fs/promises';
import { fileURLToPath } from 'url';
import { dirname, join } from 'path';

const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);

export class IndustryMatcher {
  constructor() {
    this.industryData = null;
    this.initialized = false;
  }

  async initialize() {
    if (this.initialized) return;

    try {
      const dataPath = join(__dirname, '..', 'data', 'industry-data.json');
      const data = await readFile(dataPath, 'utf-8');
      this.industryData = JSON.parse(data);
      this.initialized = true;
    } catch (error) {
      console.error('Failed to load industry data:', error);
      // Use a simplified fallback dataset
      this.industryData = this.getFallbackData();
      this.initialized = true;
    }
  }

  async classify(companyInfo) {
    await this.initialize();

    const { description, website, products, industryHints } = companyInfo;
    
    // Combine all text for analysis
    const fullText = [
      description || '',
      ...(products || []),
      ...(industryHints || [])
    ].join(' ').toLowerCase();

    // Score each industry based on keyword matches
    const industryScores = {};
    
    for (const industry of this.industryData.industries) {
      let score = 0;
      
      // Check primary keywords
      industry.keywords?.forEach(keyword => {
        if (fullText.includes(keyword.toLowerCase())) {
          score += 3;
        }
      });

      // Check secondary keywords
      industry.secondaryKeywords?.forEach(keyword => {
        if (fullText.includes(keyword.toLowerCase())) {
          score += 1;
        }
      });

      // Check industry description match
      if (fullText.includes(industry.primaryIndustry.toLowerCase())) {
        score += 5;
      }

      if (score > 0) {
        industryScores[industry.primaryIndustry] = score;
      }
    }

    // Find best match
    let bestMatch = null;
    let highestScore = 0;

    for (const [industry, score] of Object.entries(industryScores)) {
      if (score > highestScore) {
        highestScore = score;
        bestMatch = industry;
      }
    }

    // Get full classification details
    const classification = this.getClassificationDetails(bestMatch);
    
    return {
      ...classification,
      confidence: this.calculateConfidence(highestScore),
      alternativeMatches: this.getAlternativeMatches(industryScores, bestMatch)
    };
  }

  getClassificationDetails(primaryIndustry) {
    const industry = this.industryData.industries.find(
      ind => ind.primaryIndustry === primaryIndustry
    );

    if (!industry) {
      return this.getDefaultClassification();
    }

    // Get industry risk score
    const industryRisk = this.calculateIndustryRisk(industry);

    return {
      sector: industry.sector,
      industryGroup: industry.industryGroup,
      primaryIndustry: industry.primaryIndustry,
      description: industry.description,
      cpSector: industry.cpSector,
      cpIndustryGroup: industry.cpIndustryGroup,
      defaultCPGP: this.mapCPSectorToCPGP(industry.cpSector),
      industryRisk: industryRisk,
      typicalCharacteristics: industry.characteristics || []
    };
  }

  calculateIndustryRisk(industry) {
    // Map industries to risk scores based on cyclicality
    const riskMappings = {
      'Pharmaceuticals': 5,
      'Branded nondurables': 5,
      'Health care equipment': 5,
      'Telecommunications': 5,
      'Regulated utilities': 4,
      'Environmental services': 4,
      'Healthcare services': 4,
      'Retail': 4,
      'Media and entertainment': 4,
      'Technology software': 4,
      'Capital goods': 3,
      'Technology hardware': 3,
      'Chemicals': 3,
      'Building materials': 3,
      'Business services': 3,
      'Oil and gas refining': 2,
      'Homebuilders': 2,
      'Metals and mining': 1,
      'Auto OEM': 1,
      'Transportation cyclical': 1
    };

    // Check for exact matches
    for (const [key, risk] of Object.entries(riskMappings)) {
      if (industry.cpIndustryGroup?.toLowerCase().includes(key.toLowerCase()) ||
          industry.primaryIndustry?.toLowerCase().includes(key.toLowerCase())) {
        return {
          score: risk,
          riskLevel: this.getRiskLevelName(risk),
          cyclicality: this.getCyclicalityLevel(risk),
          ebitdaDeclineRange: this.getEBITDADeclineRange(risk)
        };
      }
    }

    // Default to intermediate risk
    return {
      score: 3,
      riskLevel: 'Intermediate Risk',
      cyclicality: 'Moderate',
      ebitdaDeclineRange: '10-19.9%'
    };
  }

  mapCPSectorToCPGP(cpSector) {
    const mapping = {
      'Services & Product Focus': 'Services & Product Focus',
      'Product Focus/Scale Driven': 'Product Focus/Scale Driven',
      'Capital or Asset Focus': 'Capital or Asset Focus',
      'Commodity Focus/Cost Driven': 'Commodity Focus/Cost Driven',
      'Commodity Focus/Scale Driven': 'Commodity Focus/Scale Driven',
      'Service & Product Focus': 'Services & Product Focus', // Handle variant
      'National Industries and Utilities': 'Capital or Asset Focus' // Default for utilities
    };

    return mapping[cpSector] || 'Services & Product Focus';
  }

  getRiskLevelName(score) {
    const levels = {
      5: 'Very Low Risk',
      4: 'Low Risk',
      3: 'Intermediate Risk',
      2: 'Moderately High Risk',
      1: 'High Risk'
    };
    return levels[score] || 'Unknown';
  }

  getCyclicalityLevel(score) {
    const levels = {
      5: 'Very Low',
      4: 'Low',
      3: 'Moderate',
      2: 'High',
      1: 'Very High'
    };
    return levels[score] || 'Unknown';
  }

  getEBITDADeclineRange(score) {
    const ranges = {
      5: '<5%',
      4: '5-9.9%',
      3: '10-19.9%',
      2: '20-29.9%',
      1: '≥30%'
    };
    return ranges[score] || 'Unknown';
  }

  calculateConfidence(score) {
    if (score >= 10) return 'High';
    if (score >= 5) return 'Medium';
    if (score > 0) return 'Low';
    return 'No Match';
  }

  getAlternativeMatches(scores, primaryMatch) {
    return Object.entries(scores)
      .filter(([industry]) => industry !== primaryMatch)
      .sort(([, a], [, b]) => b - a)
      .slice(0, 3)
      .map(([industry, score]) => ({
        industry,
        score,
        classification: this.getClassificationDetails(industry)
      }));
  }

  getDefaultClassification() {
    return {
      sector: 'Unknown',
      industryGroup: 'Unknown',
      primaryIndustry: 'Business Services (Default)',
      description: 'Unable to determine specific industry',
      cpSector: 'Services & Product Focus',
      cpIndustryGroup: 'Business and consumer services',
      defaultCPGP: 'Services & Product Focus',
      industryRisk: {
        score: 3,
        riskLevel: 'Intermediate Risk',
        cyclicality: 'Moderate',
        ebitdaDeclineRange: '10-19.9%'
      },
      typicalCharacteristics: []
    };
  }

  getFallbackData() {
    // Simplified industry data for fallback
    return {
      industries: [
        {
          sector: 'Technology',
          industryGroup: 'Software and Services',
          primaryIndustry: 'Software',
          description: 'Companies developing and producing software',
          cpSector: 'Services & Product Focus',
          cpIndustryGroup: 'Technology software and services',
          keywords: ['software', 'saas', 'application', 'platform', 'cloud'],
          secondaryKeywords: ['technology', 'digital', 'solution']
        },
        {
          sector: 'Consumer Discretionary',
          industryGroup: 'Consumer Services',
          primaryIndustry: 'Restaurants',
          description: 'Owners and operators of restaurants and food service',
          cpSector: 'Services & Product Focus',
          cpIndustryGroup: 'Retail and restaurants',
          keywords: ['restaurant', 'food service', 'dining', 'catering'],
          secondaryKeywords: ['hospitality', 'food', 'beverage']
        },
        {
          sector: 'Industrials',
          industryGroup: 'Capital Goods',
          primaryIndustry: 'Industrial Machinery',
          description: 'Manufacturers of industrial equipment',
          cpSector: 'Capital or Asset Focus',
          cpIndustryGroup: 'Capital goods',
          keywords: ['manufacturing', 'industrial', 'machinery', 'equipment'],
          secondaryKeywords: ['production', 'factory', 'assembly']
        },
        {
          sector: 'Health Care',
          industryGroup: 'Health Care Equipment and Services',
          primaryIndustry: 'Health Care Services',
          description: 'Providers of healthcare services',
          cpSector: 'Services & Product Focus',
          cpIndustryGroup: 'Health care services',
          keywords: ['healthcare', 'medical', 'clinic', 'hospital', 'health'],
          secondaryKeywords: ['patient', 'care', 'treatment']
        },
        {
          sector: 'Real Estate',
          industryGroup: 'Real Estate Management and Development',
          primaryIndustry: 'Real Estate Services',
          description: 'Real estate agents and brokers',
          cpSector: 'Services & Product Focus',
          cpIndustryGroup: 'Business and consumer services',
          keywords: ['real estate', 'property', 'realty', 'brokerage'],
          secondaryKeywords: ['commercial', 'residential', 'leasing']
        }
      ]
    };
  }
}