Data and Code for: Arabic Text Preprocessing for Dynamic Searchable Symmetric Encryption: An Empirical Evaluation and Preprocessor Selection Framework

Experimental result data (CSV files), relevance annotation judgments, benchmark query set, and benchmark source code supporting the findings of the paper. Includes search latency measurements for five Arabic preprocessing pipelines (NoPreprocessing, Khoja stemmer, ISRI stemmer, Lucene Arabic Analyzer, and Farasa segmenter) on the Khaleej Arabic news corpus (1,400 documents, 56 benchmark queries, 5 timed runs per query, 280 measurements per preprocessor). Also includes the annotation pool and relevance judgments for the 3,325 document-query pairs used in the search quality evaluation (inter-annotator κ = 0.956).