Similarity Measure

A Similarity Measure is a distance measure of a similarity relationship (to quantify the degree of similarity between comparable entities).

• AKA: Affinity Measure, Relatedness Function.
• Context:
  • It can enable Similarity Quantification through mathematical functions.
  • It can support Pattern Recognition through distance calculations.
  • It can facilitate Object Classification through similarity scoring.
  • It can maintain Consistency Checks through similarity thresholds.
  • It can establish Ranking Systems through similarity orders.
  • ...
  • It can often be based on a Distance Function through metric conversion.
  • It can often involve Feature Comparison through property analysis.
  • It can often promote Cluster Formation through similarity grouping.
  • It can often enhance Search Efficiency through similarity matching.
  • ...
  • It can range from being a Symmetric Similarity Measure to being a Non-Symmetric Similarity Measure, depending on its directional property.
  • It can range from being a Syntactic Similarity Measure to being a Semantic Similarity Measure, depending on its comparison level.
  • It can range from being a Simple Similarity Measure to being a Complex Similarity Measure, depending on its computation complexity.
  • ...
  • It can be computed by a Similarity Measure Computation Operation through algorithmic processes.
  • It can have a range of Similarity Scores for comparison values.
  • ...
• Examples:
  • Measure Types, such as:
    • Set Similarity for set comparisons.
    • Multiset Similarity for multiset comparisons.
    • Graph-based Similarity for graph structures.
    • Metric-Space Similarity for metric spaces.
  • Domain-Specific Types, such as:
    • Semantic Similarity Measure for meaning comparisons.
    • Word Semantic Similarity Measure for lexical comparisons.
    • Normalized Google Similarity Measure for web-based similarity.
  • Algorithm Types, such as:
    • Cosine Similarity Measure for vector comparisons.
    • Kullback-Leibler Similarity Measure for distribution comparisons.
    • Graph Nodes Similarity Measure for node comparisons.
  • ...
• Counter-Examples:
  • Dissimilarity Measure, which quantifies differences rather than similarity.
  • Distance Metric, which measures absolute separation rather than relative similarity.
  • Binary Classifier, which determines category membership rather than degree of similarity.
• See: Similarity Matrix, Clustering Task, Local Search, Equivalence Relationship, Pattern Recognition, Classification System.

---

---

References

• http://scholar.google.com/scholar?q=%22a+Similarity+Measure%22
• http://scholar.google.com/scholar?q=%22similarity+measure+such+as%22

---