The Reputation Evaluation Based on Optimized Hidden Markov Model in E-Commerce

Algorithm 1

Hidden Markov Model Algorithm based on Particle Swarm Optimization.

Input: Observation sequence, the number of hidden states

Initialization: initialized 𝑃 particles 𝑥_𝑖 and corresponding velocities 𝑣_𝑖

randomly

while a termination criterion is false do

  Compute the fitness values for all particles in swarm with (18)

  Find the local optimal solution (𝑝𝑏𝑒𝑠𝑡)

  Find the global optimal solution (𝑔𝑏𝑒𝑠𝑡)

  Update all velocities with (16)

  Update all particles with (17)

  Re-mapping particles with (20)

  Adjust the velocities of the re-mapped particles with (21)

  Re-normalization particles with (22)

end

Output: optimized model parameters 𝜆={𝜋,𝐴,𝐵}.