Coreference resolution (CR) is a current problem in natural language processing (NLP) research and it is a key task in applications such as question answering, text summarization and information extraction for which text understanding is of crucial importance. This paper describes a work in progress for improving Latvian coreference resolution that includes further experiments with the rule based LVCoref system, enlarging existing coreference corpus and the first efforts to adapt machine learning methods. LVCoref system now reaches 58.0% F-score using predicted mentions and 76.5% F-score if gold entity mentions are used.