Reading content of the Web is increasingly popular. When students read the same material, each student has a unique comprehension of the text and requires individual support from appropriate references. Most references in typical web learning systems are unorganized. Students are often required to disrupt their reading to locate references. This disruption in reading reduces learning performance. This work presents a web-based e-book learning system with an incorporated contextual knowledge recommender to assist students reading online and solve problems promptly. In this system, students can easily summarize, annotate, or enter queries in a text where questions arise. Based on a student's learning status and queries about an e-book, this system can recommend adaptive references from a knowledge repository, and locate capable classmates to answer a question. Student learning status can be analyzed based on annotations and portfolios. This knowledge repository comprises a web dictionary, discussion forum, and library of examples. Experimental results showed that students prefer accessing knowledge and joining discussions through this system to reading a conventional textbook. Most students were very willing to use this system to learn material and prepare for examinations. Furthermore, a positive correlation existed between amount of time spent using this system and exam grade. (Contains 4 figures and 1 table.)