BCCB Matrimonial — Wedding Engine

I Phase 1 — Foundation Identity Verification & Profile Authentication Signals 1–26

#	Signal	Source	Category	Weight	Implementation
01	Full legal name Extracted from government-issued ID; master name for all cross-platform matching and verification	Gov. ID	Identity	High	LLM
02	Date of birth — verified Confirmed from government ID; cross-checked against profile stated age; discrepancy over 3 years is a flag	Gov. ID	Identity	High	LLM
03	ID number format and checksum validity Country/province-specific format regex and checksum logic; invalid format = immediate hard flag	Gov. ID	Identity	High	Logic
04	ID expiry validity Expired government ID is an immediate critical flag; document must be currently valid at time of submission	Gov. ID	Identity	High	Logic
05	Document tampering — government ID Font inconsistency, pixel halos, hologram region anomaly, edge distortion, altered field detection	Gov. ID	Fraud	High	LLM
06	Face liveness detection Selfie vs ID photo biometric comparison with liveness check; confirms person submitting is the genuine ID holder	Selfie + Gov. ID	Identity	High	ML
07	Deepfake detection on profile photos GAN artifact pattern analysis on all submitted profile photos; synthetic face probability scored per image	Profile Photos	Fraud	High	ML
08	Stolen photo detection Reverse image search on profile photos; photos found on unrelated accounts or stock image sites are flagged	Profile Photos	Fraud	High	Ext API
09	Profile name vs legal name match Display name fuzzy-matched against legal name; significantly different names without a known alias pattern flagged	Profile + Gov. ID	Identity	Medium	Logic
10	Stated age vs verified age delta Profile-stated age vs ID-verified age; more than 3 years discrepancy is flagged as misrepresentation	Profile + Gov. ID	Identity	High	Logic
11	Phone number uniqueness Mobile number must be associated with only one active account; shared numbers indicate multi-account or bot activity	Registration	Identity	High	Logic
12	Email domain plausibility Temporary or disposable email domains flagged; single-use email services are a fraud onboarding signal	Registration	Identity	Medium	Logic
13	Device fingerprint uniqueness Same device used for multiple distinct identity accounts is a strong fake profile or bot signal	Behavioral	Fraud	High	Graph
14	IP geolocation vs stated location Significant mismatch between registration IP location and stated city; persistent VPN use flagged as soft signal	Behavioral	Fraud	Medium	Ext API
15	Social media account linkage Optional linkage to LinkedIn / Instagram; linked accounts with age and activity increase identity confidence	Ext API	Identity	Medium	Ext API
16	Social account age and activity plausibility Linked social accounts less than 3 months old with low activity are soft fraud signals for fake identity	Ext API	Fraud	Medium	Ext API
17	Profile photo vs social media photo match Biometric comparison of profile photos against linked social accounts confirms identity consistency	Profile + Social	Identity	High	ML
18	Criminal record / sex offender registry check Name + DOB cross-referenced against publicly available criminal and sex offender registries by jurisdiction	Ext API	Fraud	High	Ext API
19	Marital status self-declaration User-declared status: single / divorced / widowed; cross-checked against available public records where accessible	Profile	Identity	High	Logic
20	Marital status vs public record check Active marriage registration not dissolved, contradicting a "single" declaration, is a critical integrity flag	Ext API	Fraud	High	Ext API
21	Education credential stated Highest education level self-reported; used in compatibility scoring but not independently verified in phase 1	Profile	Identity	Low	Logic
22	Occupation and employer stated Self-declared occupation and employer; used for socioeconomic context and compatibility; verified in phase 2	Profile	Identity	Low	Logic
23	Location — city and region plausibility Stated city confirmed against IP, phone area code, and postal code; implausible three-way mismatches flagged	Profile + Ext API	Identity	Medium	Ext API
24	Account velocity check Multiple accounts from same device, IP, or phone within 30 days is a fraud ring or bot operation signal	Graph	Fraud	High	Graph
25	Previously banned identity detection Name + DOB + ID number cross-checked against platform ban list and partner network ban lists	Graph	Fraud	High	Graph
26	Profile completeness score Ratio of completed profile fields; low completion reduces compatibility signal quality and match confidence score	Profile	Identity	Medium	Logic

II Phase 2 — Depth Personality Profiling, Lifestyle & Values Assessment Signals 27–58

#	Signal	Source	Category	Weight	Implementation
27	Big Five personality scores (OCEAN) Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism via validated 60-item psychometric questionnaire	Psychometric	Personality	High	Psycho
28	Attachment style classification Secure / anxious / avoidant / disorganized — derived from validated ECR-R questionnaire; strongly predictive of relationship outcomes	Psychometric	Personality	High	Psycho
29	Love language preference Primary and secondary love languages: words of affirmation, quality time, acts of service, physical gifts, physical touch	Questionnaire	Personality	High	Psycho
30	Conflict resolution style Collaborative / avoidant / competitive / accommodating / compromising — assessed via scenario-based questions	Questionnaire	Personality	High	Psycho
31	Communication style preference Direct vs indirect, verbal vs written, frequency of check-ins expected; scored on validated communication scale	Questionnaire	Personality	High	Psycho
32	Emotional intelligence score (EQ) Self-awareness, empathy, emotional regulation, social skills — derived from 40-item validated EI assessment battery	Psychometric	Personality	High	Psycho
33	Introversion / extraversion energy preference How does the person recharge? Solo time vs social time balance scored on a continuous scale, not a binary category	Questionnaire	Personality	Medium	Psycho
34	Humor style and playfulness index Affiliative / self-enhancing / aggressive / self-defeating; playfulness score is correlated with relationship satisfaction in research	Questionnaire	Personality	Medium	Psycho
35	Physical activity level and preferences Gym, outdoor sports, yoga, sedentary — frequency and intensity scored; used for day-to-day lifestyle compatibility	Questionnaire	Lifestyle	Medium	Logic
36	Dietary preferences and restrictions Vegan / vegetarian / halal / kosher / omnivore; strict vs flexible; important for shared meal and social compatibility	Questionnaire	Lifestyle	Medium	Logic
37	Alcohol and substance use stance Non-drinker / social / regular / heavy; teetotaler matched with heavy drinker is a high-friction daily-life pairing	Questionnaire	Lifestyle	High	Logic
38	Smoking status Smoker / non-smoker / occasional; common dealbreaker; hard-filtered when both parties mark it as a dealbreaker	Profile	Lifestyle	High	Logic
39	Sleep schedule and chronotype Early bird vs night owl; extreme chronotype mismatches are correlated with relationship stress in longitudinal studies	Questionnaire	Lifestyle	Medium	Logic
40	Neatness and living space standards Tidiness preferences on a 5-point scale; critical for cohabitation compatibility; strong mismatch is a daily friction point	Questionnaire	Lifestyle	Medium	Logic
41	Travel frequency and style Backpacker vs luxury hotel; domestic vs international; frequency per year; adventure vs relaxation preference	Questionnaire	Lifestyle	Medium	Logic
42	Social life preferences Frequency of going out, hosting vs attending, party crowd vs quiet gatherings; social energy level alignment	Questionnaire	Lifestyle	Medium	Logic
43	Pet ownership and stance Owns pets / wants pets / allergic / dislikes pets; pet allergies and strong aversions are hard compatibility blockers	Profile	Lifestyle	Medium	Logic
44	Living arrangement intention Wants to move in soon / prefers living apart together / undecided; cohabitation timeline alignment scored	Questionnaire	Lifestyle	High	Logic
45	Children — desire and timeline Wants children / doesn't / open; preferred number and timeline; the most common long-term relationship dealbreaker category	Questionnaire	Values	High	Logic
46	Religious beliefs and practice level Faith tradition and practice intensity — practicing / cultural / non-practicing / atheist; shared faith importance weighted	Questionnaire	Values	High	Logic
47	Political orientation Self-placed on political spectrum; used for optional compatibility filtering with explicit user consent; never shown to others	Questionnaire	Values	Medium	Logic
48	Financial values and spending philosophy Saver vs spender, financial risk tolerance, views on shared finances; money conflict is a leading relationship stress driver	Questionnaire	Values	High	Logic
49	Career ambition and work-life balance stance Career-first vs family-first vs balanced; ambition alignment reduces long-term resentment in committed relationships	Questionnaire	Values	High	Logic
50	Family involvement expectations Role of extended family in couple's life; visits, holidays, financial entanglement; critical for cultural compatibility	Questionnaire	Values	High	Logic
51	Relationship type and exclusivity Monogamous / ethically non-monogamous / undecided; hard-filtered to prevent mismatched relationship structure expectations	Profile	Values	High	Logic
52	Relationship intention — seriousness Casual dating / long-term partnership / marriage-minded; intention alignment is the most basic compatibility hard filter	Profile	Values	High	Logic
53	Hobbies and interest overlap Free-text hobbies parsed and taxonomized by LLM; shared interests increase early chemistry probability and first-date ease	Profile + LLM	Lifestyle	Medium	LLM
54	Bio writing style and emotional tone analysis LLM analysis of free-text bio for personality signals, emotional tone, red-flag language, and authenticity vs template copying	Profile + LLM	Personality	Medium	LLM
55	Geographic flexibility for relocation Willing to relocate / prefers local matches / open to long-distance; location rigidity significantly shrinks compatible pool	Questionnaire	Values	Medium	Logic
56	Mental health and therapy openness Attitude toward therapy, mental health conversation, and chronic illness disclosure; openness scored for long-term compatibility	Questionnaire	Values	Medium	Logic
57	Cultural and ethnic background Self-declared cultural background; used only when user enables cultural compatibility preference; never applied as a filter by default	Profile	Values	Low	Logic
58	Love style profile (Eros, Storge, Pragma, Ludus, Mania, Agape) Lee's six love styles assessed via questionnaire; pragma and eros pairing dynamics modeled against longitudinal couple data	Psychometric	Lifestyle	Medium	Psycho

III Phase 3 — Intelligence Compatibility Engine, Behavioral Signals & Fraud Detection Signals 59–82

#	Signal	Source	Category	Weight	Implementation
59	Big Five personality compatibility delta Pair-wise OCEAN score comparison; high conscientiousness and agreeableness overlap predicts long-term relationship stability	Cross-profile	Compatibility	High	ML
60	Attachment style pairing score Secure + secure = highest probability; anxious + avoidant = high risk; modeled against longitudinal relationship outcome data	Cross-profile	Compatibility	High	ML
61	Love language compatibility score Overlap between giving and receiving love languages; mismatch between what one gives and what the other needs predicts friction	Cross-profile	Compatibility	High	Logic
62	Values alignment composite score Composite score across children, religion, finances, family, career; misalignment on any single axis is heavily penalized	Cross-profile	Compatibility	High	ML
63	Conflict style compatibility Two avoidants suppress conflict until it explodes; collaborative + competitive pairing is productive; outcome-modeled	Cross-profile	Compatibility	High	ML
64	Communication style compatibility Direct communicator paired with highly indirect; high-frequency vs low-frequency check-in preference mismatch flagged	Cross-profile	Compatibility	High	Logic
65	Lifestyle compatibility composite score Composite across sleep, activity, social, dietary, pet, tidiness signals; day-to-day friction predictor	Cross-profile	Compatibility	High	ML
66	Physical attraction preference alignment Stated physical preferences vs match profile; mutual stated attraction factors weighted symmetrically in both directions	Cross-profile	Compatibility	Medium	Logic
67	Dealbreaker hard-filter pass User-declared dealbreakers (smoking, children, religion) checked against match; any triggered dealbreaker removes the match entirely	Cross-profile	Compatibility	High	Logic
68	Socioeconomic compatibility index Income bracket, financial goals, and lifestyle spend level compared; extremes noted for user awareness, not automatically penalized	Cross-profile	Compatibility	Medium	Logic
69	Age gap and preference mutual alignment Both users' age range preferences must mutually include each other; one-directional age range matches are downweighted	Cross-profile	Compatibility	Medium	Logic
70	Response time pattern and ghosting signals Average message response time and variance; consistent ghosting patterns negatively affect user reliability score over time	Behavioral	Compatibility	Medium	Behav
71	Message sentiment and tone analysis LLM analysis of conversation tone across time; early negativity, manipulation language, or love-bombing patterns flagged	Chat + LLM	Compatibility	High	LLM + ML
72	Romance scam pattern detection LLM flags known romance scam escalation patterns: fast emotional intensity, money requests, offshore travel backstory	Chat + LLM	Fraud	High	LLM + ML
73	Financial solicitation detection Any message containing money transfer requests, gift card asks, or investment pitches triggers immediate safety review	Chat + LLM	Fraud	High	LLM
74	Catfishing behavior pattern User matches many but never meets; consistent video call refusals; inconsistent personal story detected across messages by LLM	Behavioral + LLM	Fraud	High	ML + LLM
75	Harassment and abusive language detection Real-time message moderation for explicit abuse, sexual harassment, coercive language, and threatening content across all conversations	Chat + ML	Fraud	High	ML
76	Match-and-harvest pattern detection Users with unusually high match rates who never progress to dates or conversations flagged for automated behavior review	Behavioral	Fraud	Medium	Logic
77	Report and block frequency Users frequently reported by other members have trust score progressively reduced; pattern of reports escalates to human review	Graph	Fraud	High	Graph
78	Video call verification completion Users who complete a video call before first date receive a "video verified" badge; significantly raises trust score	Platform	Compatibility	Medium	Logic
79	Date confirmation and outcome feedback Users confirm dates occurred and optionally submit outcome; this feedback data continuously trains the compatibility ML model	Platform	Compatibility	High	Graph
80	Network fraud ring detection Graph clustering on shared device, IP, photo hash, or template attributes identifies coordinated fake profile networks	Graph	Fraud	High	Graph + ML
81	Mutual social connection graph Shared social connections via linked accounts; mutual friends positively affect early trust and conversation safety perception	Social Graph	Compatibility	Low	Graph
82	Swipe behavior revealed preference analysis Implicit preference signals from swipe history refine explicit stated preferences; stated preference ≠ revealed preference detection	Behavioral	Compatibility	Medium	ML

IV Phase 4 — Output Score Aggregation, Explainability & Trust Layer Signals 83–92

#	Signal	Source	Category	Weight	Implementation
83	Weighted compatibility score (0–100) Composite score across all categories with research-calibrated weighting; values and attachment signals weighted highest	Aggregate	Meta	High	ML
84	Identity trust score (0–100) Separate score reflecting depth of identity verification; surfaced as a badge on profile visible to all potential matches	Aggregate	Meta	High	Logic
85	Compatibility category sub-scores Separate scores for Personality, Lifestyle, Values — shown to both matches with mutual consent before or after conversation	Aggregate	Meta	Medium	Logic
86	Plain-English compatibility narrative LLM-generated match summary highlighting strengths and gentle discussion points: "You both prioritize family but differ on city vs suburbs"	Aggregate + LLM	Meta	High	LLM
87	Red flag auto-escalation Any single critical flag — fake ID, romance scam pattern, criminal match, financial solicitation — triggers immediate safety REVIEW	Aggregate	Meta	High	Logic
88	AI-generated conversation starters Icebreaker questions generated based on shared interests and compatibility gaps; reduces opening message anxiety and rejection	Aggregate + LLM	Meta	Low	LLM
89	Model confidence score How confident is the compatibility prediction given profile data completeness; low-confidence matches are flagged for users	Aggregate	Meta	Medium	ML
90	Network safety annotation If either profile has network-level fraud signals, a private safety advisory is surfaced to the other user before engagement	Graph	Meta	High	Graph
91	Compatibility model continuous feedback loop Date outcomes, relationship reports, and ghosting data continuously retrain the ML model to improve match quality over time	Graph + ML	Meta	High	Graph + ML
92	Portable verified identity certificate Fully verified users receive a shareable cross-platform trust credential; builds network moat and reduces re-verification friction	Graph	Meta	High	Graph + ML

Identity Verification & Compatibility Engine