The present invention relates to a spammer group extraction apparatus and method, which extract spammer groups that interfere with fair trade and unbiased decision making by sending messages aimed at intentionally slandering other companies (other persons, other products, etc.) on social network services. The spammer group extraction apparatus includes a data collection unit for collecting pieces of data corresponding to social network services. A natural language processing unit preprocesses the pieces of data using a natural language processing algorithm based on big data. An abnormal behavior detection unit detects abnormal behavior based on user identifications (IDs) respectively corresponding to pieces of data, preprocessing of which has been completed. A spammer extraction unit extracts a spammer group using a user ID causing the abnormal behavior and an ID of a user group including the user ID.