In　this　paper,　inspired　by　a　unified　probabilistic　model　of　information　retrieval,　we　propose　a　general　framework　of　targeted　marketing　by　considering　three　types　of　information,　namely,　the　customer　profiles,　the　product　profiles,　and　the　transaction　databases.　The　notion　of　market　value　functions　is　introduced,　which　measure　the　potential　value　or　profit　of　marketing　a　product　to　a　customer.　Four　sub-models　are　examined　for　the　estimation　of　a　market　value　function.　Based　on　market　value　functions,　two　targeted　marketing　strategies,　namely,　customer-oriented　targeted　marketing　and　product-oriented　targeted　marketing,　are　suggested.　This　paper　focuses　on　the　conceptual　development　of　the　framework.　The　detailed　computation　of　a　market　value　function　and　the　evaluation　of　the　proposed　framework　will　be　reported　in　another　paper.　©　Springer-Verlag　Berlin　Heidelberg　2005.