This paper describes our approach to applying type-oriented inductive logic programming (ILP) to information extraction (IE) tasks and the latest experimental results in learning IE rules from the data generated from 100 newspaper articles. Information extraction involves extracting key information from text corpus in order to fill empty slots of given templates. A bottle neck in building IE systems is that constructing and verifying IE rules is labor intensive and time consuming. To automatically generate IE rules, we employ an ILP system RHB+ that learns logic programs whose variables have type information. After giving our approach to applying ILP to information extraction tasks in detail, the process of learning IE rules is illustrated. Experiments were conducted on the data generated from 100 newspaper articles relating to release of new products. The results show high accuracy and precision of the learned rules. This indicate that type-oriented ILP has a high potential for the use of automatically generating IE rules.