Tidal interactions are a key process to understand the evolution history of close-in exoplanets. But tidals still have a large uncertainty in their prediction for the damping timescales of stellar obliquity and semi-major axis. We have worked on a search for transiting giant planets around evolved stars, for which few close-in planets were discovered. It has been reported that evolved stars lack close-in planets, which is often attributed to the tidal evolution and/or engulfment of close-in planets by the hosts. Meanwhile, Kepler has detected a certain fraction of transiting planet candidates around evolved stars. Confirming the planetary nature for these candidates is especially important since the comparison between the occurrence rates of close-in planets around main sequence stars and evolved stars provides a unique opportunity to discuss the final stage of close-in planets. With the aim of confirming KOI planet candidates around evolved stars, we measured precision radial velocities (RVs) for evolved stars with transiting planet candidates using Subaru/HDS. We also developed a new code which simultaneously models and fits the observed RVs and phase-curve variations in the Kepler data (e.g., transits, stellar ellipsoidal variations, and planet emission/reflected light). As a result of applying the global fit to KOI giants/subgiants, we confirmed two giant planets around evolved stars (Kepler-91 and KOI-1894), as well as revealed that KOI-977 is more likely a false positive.