This study tries to detect the impact research topics from impact authors with their connections, that is, who have larger impact in the same research field. These topics are impact research topics the pursuit of which would be very valuable for researchers, especially for new scholars or for researchers who want to combine their original field with other new domains but who may not have enough background knowledge about the new field. Bayesian estimation in our model uses subjective data (published volume) as the prior distribution and objective data as the likelihood function (citation frequency) to predict the posterior distribution of the target which we called impact power. After finding the impact power of each paper or topic then filtering these papers and topics, we can find impact research topics or papers.