mssrchapelier/tigre-parser

A rule-based morphological analyser for Tigre (Eritrea) in Java (bachelor's thesis - Moscow State Univ., 2019, refactored 2020). # TigreParser TigreParser is a morphological analyser of the Tigre language (an Ethiosemitic language spoken by around 1 million people, most of whom live in Eritrea), written in Java. This analyser was first developed in 2019 as part of a bachelor's thesis at (Lomonosov) Moscow State University, Russia (Department of Theoretical and Applied Linguistics, Faculty of Philology) [ISTINA ref. - MSU]. The thesis itself, which describes the first version of the application, is available here (in Russian only). The architecture has since been changed to quite a drastic extent, mainly to allow for greater modularity, but the basic algorithm has stayed the same. ## How to use ### Trying out / using without customisation 1. Download and install the Java Runtime Environment for your operating system. 2. Download `tigreparser-bundle.jar` and run it as follows: ``` java -jar tigreparser-bundle.jar input_file ``` You can specify the desired **output file name** with the `-o` / `--output` flag (if left unspecified, the output will be written to `output_input-file-name` in the same folder as the `.jar` file): ``` java -jar tigreparser-bundle.jar input_file -o output_file ``` or ``` java -jar tigreparser-bundle.jar input_file --output output_file ``` By default, the output will include both complete (if any) and incomplete analyses; complete analyses are shown first, then the incomplete ones (the more morphemes, the closer to the top of the list). This may make for an excessive number of analyses to be put into the output file; you can set a **cutoff value** (a positive integer) with the `-n` / `--numanalyses` flag to set the maximum number of analyses that will be returned for each word: ``` java -jar tigreparser-bundle.jar input_file -n 5 ``` or ``` java -jar tigreparser-bundle.jar input_file --numanalyses 5 ``` A reasonable value for this parameter seems to be between 3 and 5. The flags can be combined (but none are required; the only parameter you must pass to the prog …