Emotion cause is an essential feature in generating empathetic responses. The process of finding the emotional causes for a particular is emotion cause extraction. This work proposes a multi-attention-based approach that utilizes textual and numerical features to enhance the cause extraction task. We target the usage of moderate-size datasets to show the efficiency of our model, as generally, the conversational datasets in English are of the mentioned size. In our work, we prove the efficiency of our model on the RECCON dataset against various state-of-the-art models and notice a significant amount of improvement.