Talk:SimpleGraph-Excel

Mental Model

Mapping rules

1. Each Vertex type (by label) is converted to an Excel sheet
2. Each Edge type (by label) is converted to an Excel sheet
3. Each vertex is converted to a row
4. Each vertex property is converted to a cell
5. Each edge is converted to a row
6. Each edge property is converted to a cell
7. The header row for Vertices has a column for each property key
8. The header for for Edges has a column for each property key + an in and and out column
9. The in and out header columns for edges have the label of the corresponding vertex in parentheses

Extended Mapping Rules

Range Mapping

1. Ranges can represent collections of vertices or edges
2. Named ranges map to specific vertex/edge label types
3. Each range spans multiple cells corresponding to multiple graph elements
4. Range references in formulas map to graph traversals across multiple vertices/edges

Formula Mapping

1. Formulas stored in cells can represent computed properties
2. Formula results are mapped to property values in the graph
3. Formulas can reference other vertices/edges through cell references
4. Complex formulas map to graph traversal expressions

Reference Mapping

1. Cell references in formulas map to vertex/edge relationships
2. Absolute references ($A$1) map to fixed vertex IDs
3. Relative references (A1) map to traversal patterns
4. Cross-sheet references map to cross-label relationships between different vertex/edge types

Comment Mapping

1. Comments on cells map to metadata properties on vertices/edges
2. Comment author maps to a metadata property
3. Comment text maps to annotation content
4. Comment timestamps preserve temporal information in the graph

Formula AST Mapping

1. The Abstract Syntax Tree of formulas maps to graph operations
2. FunctionCall nodes map to graph traversal operations:
   1. SUM() → aggregate property values across connected edges
   2. VLOOKUP() → vertex lookup by property value
   3. IF() → conditional edge creation or property assignment
3. BinaryOperation nodes map to property computations or edge weights:
   1. Arithmetic operators (+, -, *, /) → computed properties from multiple vertex properties
   2. Comparison operators (=, <, >) → edge filtering conditions
4. LiteralValue nodes map to constant property values directly stored in the graph

UML diagram

Bidirectional Transformation

Excel to Graph

1. Parse workbook structure
2. Create vertex types from sheet names (non-edge sheets)
3. Create edge types from sheet names (sheets with in/out columns)
4. For each row in vertex sheets:
   1. Create vertex with label = sheet name
   2. Map each cell to a property
   3. Evaluate formulas and store results
   4. Attach comments as metadata properties
5. For each row in edge sheets:
   1. Create edge with label = sheet name
   2. Resolve in/out references to vertices
   3. Map remaining cells to edge properties
6. Process named ranges as vertex/edge collections
7. Build formula AST for computed properties

Graph to Excel

1. Create sheet for each unique vertex label
2. Create sheet for each unique edge label
3. For each vertex:
   1. Create row in corresponding sheet
   2. Create cell for each property
   3. Add comments from metadata properties
4. For each edge:
   1. Create row in corresponding sheet
   2. Add in/out columns with vertex references
   3. Create cell for each property
5. Create named ranges for logical vertex/edge groupings
6. Reconstruct formulas from computed properties with AST

Example Mapping

Graph Example

// Vertices
person1 = g.addV('Person').property('name', 'Alice').property('age', 30).next()
person2 = g.addV('Person').property('name', 'Bob').property('age', 25).next()
company1 = g.addV('Company').property('name', 'TechCorp').property('revenue', 1000000).next()

// Edges
g.addE('WORKS_AT').from(person1).to(company1).property('salary', 75000).next()
g.addE('WORKS_AT').from(person2).to(company1).property('salary', 65000).next()
g.addE('KNOWS').from(person1).to(person2).property('since', 2020).next()

Excel Example

Sheet: Person

name	age	totalSalary (Formula)
Alice	30	=VLOOKUP(A2,WORKS_AT!A:C,3,FALSE)
Bob	25	=VLOOKUP(A3,WORKS_AT!A:C,3,FALSE)

Sheet: Company

name	revenue	avgEmployeeSalary (Formula)
TechCorp	1000000	=AVERAGE(WORKS_AT!C:C)

Sheet: WORKS_AT (Edge)

out (Person)	in (Company)	salary
Alice	TechCorp	75000
Bob	TechCorp	65000

Sheet: KNOWS (Edge)

out (Person)	in (Person)	since
Alice	Bob	2020

Named Ranges:

• AllEmployees = Person!A2:B3
• CompanyFinancials = Company!A2:C2
• Salaries = WORKS_AT!C2:C3

Comments:

• Cell Person!A2: "CEO of department" (Author: Admin, Date: 2024-01-15)
• Cell Company!B2: "Annual revenue in USD" (Author: Finance, Date: 2024-01-10)

Implementation Notes

1. Formula evaluation must handle circular references between sheets
2. Cross-sheet references require maintaining vertex/edge ID mappings
3. Named ranges improve query performance by pre-defining common traversals
4. Comments preserve audit trail and metadata
5. AST parsing enables complex graph queries from Excel formulas
6. Absolute vs relative references determine query pattern (fixed vs traversal)