I looked through the web, searching for information about problems with Ithenticate and symbols or greek letter, and I couldn't find any information about how to change the bahviour of Ithenticate in such a way. You should probably just try to reduce the index through other means. To copy a comment from the now deleted user32754 so that it won't get lost in the future:
I can't quite believe that a single word is responsible for a problematic similarity index. There must be many more similarities between your text and published articles, if you reach a problematic level. Maybe if you take care of the other similarities, this one word won't matter any more. —— Also, if the similarity is obviously caused by a software problem (that is, the software is unable to differentiate θ-continuous from continuous) you should be able to bring up this problem with whoever is responsible for the evaluation of your text.
You could also try to write the word differently. Again, a comment, this time from the user fer-rum:
What happens if you write it as "theta-continuous"? Being familiar with software development, it would not surprise me if the tool just skipped everything that it can't make out to be a word (And all words are written in Latin alphabet aren't they?), leading to the exact problem you describe.
To support the above thesis I looked at what Ithenticate thinks about other letters and symbols. For example here you can find information about what Ithenticate thinks about symbols in their article Plagiarism Detection Software MISCONCEPTIONS under the heading "Plagiarism Detectors Are Easy to Deceive":
Especially as more academic institutions and publishers begin mandating the use of plagiarism detection software, there are increasing attempts to devise methods of tricking the system into passing off duplicated text as original. Students have gone to great lengths to cheat the system—substituting Cyrillic letters that look similar to Latin letters throughout the text, adding extra spaces between letters and words, or even adding “invisible” small letters in white font to differentiate the text, often posting their strategies online for others to use. What is less publicized is that behind the scenes, the creators of plagiarism detection software go to even greater lengths to anticipate cheating strategies and to incorporate fixes into their systems. Many plagiarism detection software systems automatically strip macros from submitted manuscripts and can easily detect oddly placed punctuation and invisible characters.
It's quite possible that your addition of the greek letter theta is simply recognized as an attempt to circumvent the normal rules for detecting plagiarism and there are methods in place to deal with such an attempt - namely ignoring this symbol.
It won't be easily possible to change this from your position. The exact rules won't be published anywhere. Otherwise everyone could simply get around the software and they wouldn't survive very long in the market.