Fragment-based genetic programming for fully automated multi-objective web service composition

Evolutionary Multitasking for Semantic Web Service Composition

“…For example, [6] samples permutation-based solutions from learning the distribution models of promising solutions in each generation. The second subgroup only reported two recent attempts on single-tasking multi-objective fully automated web service composition [12], [13]. For example, a hybrid approach [12] was proposed to decompose the multi-objective problem into single-objective subproblems, producing a set of Pareto solutions with trade-offs.…”

“…Existing service composition algorithms are designed primarily to solve each service composition request independently by using either single-objective [3], [8], [9], [11] or multi-objective approaches [12], [13], ignoring similarities between different requests that could be dealt with collectively. For example, many service composition requests have similar input and output requirements but may vary due to different preferences on their quality.…”

“…Service composition problem is NP-hard and cannot be solved in polynomial time [7]. Due to this reason, Evolutionary computation (EC) techniques have been proposed to efficiently find near-optimal solutions that satisfy users' requirements reasonably well [2], [3], [4], [6], [8], [9], [10], [11], [12], [13]. These EC-based service composition approaches are mainly classified into two groups based on the number of objectives to be optimized: single-objective [2], [3], [4], [6], [8], [9], [10], [11] or multi-objective approaches [12], [13].…”

“…Due to this reason, Evolutionary computation (EC) techniques have been proposed to efficiently find near-optimal solutions that satisfy users' requirements reasonably well [2], [3], [4], [6], [8], [9], [10], [11], [12], [13]. These EC-based service composition approaches are mainly classified into two groups based on the number of objectives to be optimized: single-objective [2], [3], [4], [6], [8], [9], [10], [11] or multi-objective approaches [12], [13]. The first group optimizes only one objective by combining all quality criteria into one (e.g., one combined quality that measures QoSM and QoS [2]); the second group has an aim to identify a group of composite services with varied trade-offs over multiple objectives (e.g., two trade-off objectives: one combines time and cost, the other combines availability and reliability [12]).…”

EDA-based approach to comprehensive quality-aware automated semantic web service composition

“…In particular, a group of layers is identified, and each layer contains a set of services that have the same longest distance to Start. We adopt a layer discovering method in [44] to find relevant services and service layers as illustrated in the following example. Fig.…”