Kinyarwanda Stemmer

With the goal of developing language technology for Kinyarwanda, we seek volunteers to participate in a stemming task. In order to do natural language processing (NLP) for Kinyarwanda, morphological segmentation is a fundamental task. We have developed a tool that performs morphological analysis; however, the tool is not perfect. We need your help to improve the accuracy of the tool by performing an easier task: stemming Kinyarwanda verbs.

STEMMING TASK: In each screen, you are given a conjugated verb and a list of candidate infinitives. The task is then to choose the infinitive which you think has the most original root of the given conjugation. For example, given "yabajijwe: kubaza, kubara, kubazwa, kubaga", the answer should be "kubaza". If none of the candidate infinitives is valid, then the answer should be "Nta n'imwe irimo".

LEXICAL TASK: In each screen you will be given a verb. The task is to decide whether the verb is an original known lemma, a derived or inflected verb or unknown (i.e. you don't know the meaning of the verb).

Only your answers will be collected and the Internet cost should be as low as typing a small text on WhatsApp. We very much appreciate your help in this effort. If you have any questions, concerns, suggestions or would like to collaborate further on the project, please contact us by email: anthonzeyi@gmail.com.

About the developer: Antoine Nzeyimana is a software engineer and a graduate student at the University of Oregon.

Recent changes:
STEMMING TASK: In each screen, you are given a conjugated verb and a list of candidate infinitives. The task is then to choose the infinitive which you think has the most original root of the given conjugation. For example, given "yabajijwe: kubaza, kubara, kubazwa, kubaga", the answer should be "kubaza". If none of the candidate infinitives is valid, then the answer should be "Nta n'imwe irimo".