The purpose of this thesis is to explain the logic and web dictionary, and how to refine the automatic correction of spelling orthography. In these days, so many people depend on the web dictionary much more and the web dictionary get a great roles in the life along with the change of media. The most big difference between paper dictionary and web dictionary is the function of automatic correction in spelling orthography. The automatic correction is the devices to correct the words input by the users considering user’s intention in the level of words or sentences. This thesis mainly focused on the lexical level applied in Daum web Korean dictionary. Daum web Korean Dictionary consist of lists based on the comparison of misused words, syllables, the patterns of high frequence words and phoneme analysed by initial, medial and final positions. Actually there are some problems because it has some limitations and contradiction in the respective logics. I’d like to suggest some ideas to correct and improve this problems and limitations through enlarging patterns of phonemic fluctuation, compared words with verbal inflection, eliminating unavailable words, applying the result of machine learning and improvement of misused words, etc. It is also important to show the result corrected spelling orthography so that the user can use the function more easily. At present, the service has two steps which present the word list searched by the user first and then suggest the candidate words next. In the future we need to develop the service system to present the high frequence words list of mistakes or error with additional explanation to educate the users effectively.