Pattern and Link Analysis with JanusDB and Dgraph

Pattern Analysis and Applications Overview

A pattern is composed of common characteristics forming a sequence occurring at regular intervals. Pattern Analysis involves an interpretation of the pattern to make predictions of data.

Steps in Pattern Analysis involves –

  • Pattern contains description
  • Intent and Motivation uses analysis patterns
  • External Factors affecting the pattern
  • Implementation of pattern in the analysis model
  • Consequences describe the after-effects of analysis pattern into the application
  • Design helps to achieve design patterns
  • Similar Patterns to discover related patterns

Link Analysis centers relationships and connections between network nodes. Link Analysis performs matching for known patterns, detection of anomalies, discover new patterns.

Steps in Link Analysis involves –

  • Count Total Links
  • Check Unique Domains
  • Anchor Text Variance
  • Incoming Links
  • Page performance
  • Link Quality

Challenge for Pattern Analysis

To view the relationship between the user who transacted and received the amount of transaction data.

Transaction data in the form of graph containing –

  • Type
  • Amount
  • Origin
  • Destination
  • Information about the user

Solution Offered for Pattern and Link Analysis

Graph to represent the relationship between the origin(the user who transacted to another user) and destination (the user who received the amount paid by the origin).

Tinker graph creates the vertices and edges using the CSV file containing transaction data.

The relationship between the vertices represented in the form of edges. The relationship that exists between the vertices is stored in the form of XML in the .graphml file as an output of the Tinker graph.

This file is viewed by Gephi to visualize the output in the form of the graph.

Read more