الفهرس | Only 14 pages are availabe for public view |
Abstract This thesis presents an evaluation of the Stanford constituency parser based on phrase structure theory. The main approach is to measure the accuracy of the parser in dealing with the constituents of the sentences through evaluating its output. Two hundred and fifty sentences are randomly chosen and the parser is applied to them. The process of evaluation is done through using three common metrics to evaluate the performance of constituency parsing. These metrics are precision, recall, and F1-score. The results show that the levels of accuracy obtained from the parser output vary depending on the syntactic categories of the constituents and the different structures of the sentences. The parser achieves a recall of 100% and a precision of 80%. Several training methods should be applied to the parser to enhance its performance. |