We proposed an approach to automatically predict the evaluation of consultants in press conferences using text only. The word representation consists of token embedding using ELMo and type embedding. The language model we used is an LSTM with self-attention mechanism. We collected seven publicly available press conference videos and all the Q&A pairs between speakers and journalists were annotated by professional consultants. As a result, we achieved the average accuracy of 57.62% for prediction of all evaluation criterions.