基于 Context 建模的熵编码已经获得广泛应用。实际应用中的主要问题是如何建立有效的 Context 模型，既提高熵编码效率又降低模型代价。基于Context树模型的优化与context加权是解决这个问题的有效手段。但目前这两种技术都还有改进的空间。本项目基于最小描述长度理论，研究context树模型的更细粒度优化问题；探索context加权代价和可选择加权。同时，针对基于参考的基因组压缩问题，研究序列片段的相似测度和最佳匹配策略，探索基于后缀数组的匹配序列查找算法，研究压缩数据的随机访问以及适合随机访问分组压缩的context建模问题。研究成果将以论文形式在国内外重要期刊上发表论文 10 篇以上（SCI，EI收录6-8篇）

Context modeling-based entropy coding has been applied in many applications. The major obstacle in practice is how to construct an effective context model which can improve the entropy coding efficiency as well as reduce the model cost. The optimization of the context tree model and context weighting are two effective methods to tackle the above problem. However, both of them can still be improved. In this project, the tuning of the context tree model will be performed on a more fine-grained basis. The context weighting cost and optional context weighting will be explored. Meanwhile, for the referential-based genome sequence compression, the similarity measure between the reference and target subsequences as well as the suffix array-based optimal matching strategy will be studied. The random access of the compressed sequence along with the corresponding context modeling strategy will also be studied in this project.