Skip to content

Latest commit

 

History

History
53 lines (36 loc) · 1.54 KB

README.md

File metadata and controls

53 lines (36 loc) · 1.54 KB
Raw

Game Title Comparison Algorithm

Usage

The comparison function takes two sets of game titles. A set can contain titles for a game (translations, alternative spellings, etc.).

Every entry in a set is compared to each entry in the other set. The best result (a value between 0.0 and 1.0) will be returned.

E.g.

from comparison_algorithm import cmp_titles

cmp_titles(
    ["Final Fantasy VII"],
    ["FF 7", "Final Fantasy 7"]
)

returns 1.0

Preprocessing

  • Remove trademark symbols
  • Replace Ⅱ symbol with correct roman numeral II
  • Remove brackets (with content)
  • Remove〔〕brackets
  • Remove common (Japanese) game series names from titles (e.g. "SIMPLE 1500")

Standardization

Before comparison the title strings will be standardized:

  • Remove articles ("The", ...)
  • Remove punctuation
  • Standardize Japanese macron and vowel transcription
  • Set to lower case

Rules

The comparision is based on Levenshtein distance between the two strings. Additionaly, some rules were implemented to take specific features of game titles into account.

Numbering

Using Levensthein distance, a differnt number in games title can still result in a high similarity score in the case of longer titles. This rule therefore adds a penalty weight if the numbers in the titles mismatch. Additionally, this rule considers:

  • roman and arabic numerals are treated as the same value
  • position of the number in the title

First letter

This rule adds a penalty weight if the first letter of the strings mismatch. (Only applied in case of short titles).