The construction of personalized vocabulary learning paths is the key to improving the effectiveness of English vocabulary teaching for undergraduate college students, which can increase students’ engagement and motivation, and thus enhance learning effectiveness. In this paper, the Felder-Silverman Learning Style Scale and K-Means clustering algorithm are used to represent the English vocabulary learning styles of undergraduate college students, and the learner model and learning knowledge point model are constructed based on the CELTS-11 learner model specification. Then the decision variables and objective functions of personalized learning paths were constructed from the learner characteristics and combined with English vocabulary learning resources. Considering that the Ant Colony algorithm may fall into local optimality when solving the optimization results, this paper improves the A-ACO algorithm through the heuristic function and pheromone reward and punishment mechanism, and uses it to solve the English vocabulary personalized learning path optimization model. It is found that the A-ACO algorithm has high stability and accuracy in solving the model, and the optimized learning path is used in English vocabulary practice teaching, the students’ English vocabulary scores are improved by 16.51 points as a whole, and the vocabulary richness of the students is also improved significantly. The optimization results of English vocabulary learning paths for undergraduate college students solved by Ant Colony algorithm help students to systematically and comprehensively master the lexical properties, meanings and collocations of common words, and lay a solid foundation of vocabulary for students’ English learning.
