Data for: Semi-Supervised Lexicon Generation Using Semantic Relations for Dream Content Analysis

Presented is an implementation of the SALAD algorithm for dream content analysis through word searching. Helper functions for constructing initial seed word dictionaries are provided in "hyponym_dictionary.py" which will also be used to construct the dictionaries from the seed words. "read_csv.py" reads and pre-processes the dream reports into a dictionary that captures the linguistic features of the words and sentences from the dreams. It also contains an implementation of the Improved Lesk Algorithm. The folder Series/ can be populated with data from any dream journal (you can take data from www.dreambank.net). The required data format is a csv file containing one dream in each row. The code "search_lemmas.py" performs the actual word search. The exact steps of SALAD and the parameters that need to be played around with to obtain the best results are described in the paper. The codes are written in Python 3.6 and can run on Python 3.6 and above.本项目实现了用于梦境内容分析的SALAD算法（SALAD algorithm），该算法通过词汇检索实现梦境内容解析。用于构建初始种子词汇词典的辅助函数已在`hyponym_dictionary.py`中提供，该文件同时可基于种子词汇生成对应词典。`read_csv.py`用于读取梦境报告并完成预处理，将其转换为可捕捉梦境中词汇与语句语言特征的词典格式，同时该脚本还实现了改进型莱斯克算法（Improved Lesk Algorithm）。`Series/`文件夹可填充任意梦境日记数据集，数据来源可参考`www.dreambank.net`。所需数据集格式为CSV文件，其中每一行对应一则梦境报告。`search_lemmas.py`负责执行实际的词汇检索任务。SALAD算法的具体执行步骤，以及为获取最优结果所需调试的各项参数，均已在相关论文中详述。本项目所有代码均基于Python 3.6编写，可在Python 3.6及以上版本环境中运行。