The effect of different video summarization models on the quality of video recommendation based on low-level visual features.