In daily life, people communicate by conveying their emotions, thoughts, and experiences to the people around them with or without language. Hovewer, persons who do not have speech and hearing skills make this communication with their environment by using sign language. As for that the sign language is defined as a language enabling handicapped persons communicate through gesture, mimic and body movements between themselves and with other persons who know this language. In this sense, this work developed for its functioning at the web medium through the semantic analysis models over the Internet of Things is a Turkish Sign Language Translation System which provides a communication opportunity between hearing impaired and speech handicapped persons and those who do not know the sign language. Through this system, movements of hands reflecting the sign language were captured by a depth camera and sent to the MQTT (Message Queuing Telemetry Transport) server. These images were analysed by means of the methods of three dimensional modelling, artificial neural networks algorithms used at movement analysis and machine learning. Real meanings of the words obtained from the analysed images were found through the reality models consisting of thematic roles prepared in accordance with the Formal Concept Analysis Theory. These words the real meanings of which had been identified were published at the local network medium together with the sentence that they had been mentioned. Thus, a handicapped person could communicate with other person or persons being at the same place or elsewhere.