Abstract: In a dynamic constructed business workflow,users always leave out the activities they should do or select the wrong activities they shouldn’t do at the background of complex enterprise business logic．To solve this problem,a semantic-based approach for business activities recommendation was proposed．First we defined the foundation ontology model for business activities,and on the basis of the model built some basic reasoning rules．Then we described the re-commendation strategy in detail,and designed the recommendation algorithm based on the strategy．The mechanism of the approach is receiving and analyzing business activity events to invoke the recommendation algorithm,then the algorithm accesses to knowledge base to generate the final recommended results to the end user．Finally,a case of state comprehensive disaster reduction system was studied to validate the proposed approach.

Key words: Event-driven,Semantic,Reasoning rules,Business activities,Recommendation