State-of-the-Art Review on Relevance of Genetic Algorithm to Internet Web Search

Table 1

Comparison of Diverse Proposals that use genetic algorithm for internet search.

Diverse proposals	Reason for GA	Chromosomes	Fitness function adopted	Genetic operators used
(Eissa and Alghamdi, 2005) [30]	Genetic algorithm is used to optimize the profiles whereas the relevance feedback is used to adapt it.	Represent a gene as a term, an individual as a document and the population as the profile.	𝐹(𝑃𝑖)=𝑘∑𝑆𝐷𝑃𝑘,𝑃𝑖/#𝐷𝑝.	Selection.
(Vallim and Coello, 2003) [31]	Combines user’s feedback to new documents retrieved by the agent with a genetic algorithm.	Individuals represented by a query vector and its adaptation rate.	𝑄=𝑄+𝛼𝑓; 𝐹(𝑄)=𝑓+𝛽𝑓.	Two point crossover and Mutation operator.
(Li et al., 2000) [32]	Realize the scheduling strategy of agent manager.	Search space is represented as weight field in the search engine. Field are search parameters.	Adaptation function; Φ(agent) = Γ(𝑓,𝑝,𝑐,𝑢,𝑡).	One point Cross over and Single point Mutation.
(Caramia et al., 2004) [33]	Select a subset of original pages for which the sum of scores is large.	Chromosomes represent subsets of pages of bounded cardinality. Each page is a gene.	𝐹𝑓(𝑐)=𝛼⋅𝑡1(𝐶)+𝛽⋅𝑡2(𝐶)+𝛾⋅𝑡3(𝐶).	Single point crossover.
(Rocio et al., 2008) [34]	Evolving lofty quality query.	Chromosome is represented as a list of terms where each term corresponds to a gene.	Fitness (q) = max(𝜎(𝑐,𝑑𝑖)) 𝑑𝑖𝜀𝐴𝑞.	Roulette Wheel Selection, Single point crossover, One point mutation.
(Abe et al., 1999) [35]	For evolving information retrieval agents.	Genes are represented by the search parameters.	𝐹= (SH/MH + SI/MI) ∗ (1 − ST/TL) + (1 − ME/MM).	Selection uses ranking strategy, Uniform crossover, and Single Point Mutation.
(Martin-Bautista et al., 1999) [36]	Adaptive internet information retrieval.	Each gene represents a fuzzy subset of the document set by means of a Keyword term and number of occurrences in a document.	𝑓^𝑗𝑖=𝑓𝑖^𝑗−1+𝑃^𝑗𝑖−𝐿^𝑗𝑖; where 𝑃^𝑖j and 𝐿^𝑖j is the pay off of life tax and chromosome number respectively.	Random selection, Double point Crossover, Random Mutation.
(Marghny and Ali, 2005) [19]	Steady state genetic algorithm for optimizing web search.	Initial population is generated by heuristic creation operator which queries standard engines to obtain pages.	Fitness function evaluates web pages is a mathematical formulation of Link quality, Page quality and Mean quality function.	Binary tournament selection, Single point crossover.
(Cheng et al., 1998) [22]	GA implemented as a spider to find most relevant home pages in the entire internet.	Chromosomes represent all input home pages in a set.	Jaccard’s coefficient function.	Heuristic based cross over, Simple mutation.
(Lin et al., 2002) [20]	Improving the searching performance.	Initial population represented by binary coding selected at random.	𝐹𝑑 = (𝐹max−𝐹min)/𝐹aver.	Fitness proportion selection, Adaptive adjusting crossover, Mutation operation range.
(Fan et al., 2003) [37]	Genetic Programming to the ranking function discovery problem leveraging the structural information of HTML documents.	Chromosomes represent html pages.	The fitness evaluation of each ranking tree is done at the level multiple queries.; 𝑃Avg=∑^𝑇Re𝑙𝑖=1𝑃𝑖/𝑇Re𝑙,; 𝑃𝑖=𝑖/Rank𝑖.	Single point Crossover, One point Mutation.
(Milutinovic et al., 2000) [38]	Genetic search algorithms enable intelligent and efficient internet searches.	Chromosomes represent set of input Web sites given by a user.	Jaccard’s Function.	Topic Mutation, Spatial Mutation, Temporal Mutation.
(Koorangi and Zamanifar, 2007) [14]	Query reformulation in search engine.	Initial population consists of first five keywords of the user dictionary.	CHK fitness function.	One point crossover, Inversion mutation operator.