Context awareness is one of the building blocks of many applicationsin pervasive computing. Recognizing the current context of a useror device, that is, the situation in which some action happens, oftenrequires dealing with data from different sensors, and thus differentdomains. The Growing Neural Gas algorithm is a classification algorithmespecially designed for un-supervised learning of unknown input distributions;a variation, the Lifelong Growing Neural Gas (LLGNG), is well suitedfor arbitrary long periods of learning, as its internal parametersare self-adaptive. These features are ideal for automatically classifyingsensor data to recognize user or device context. However, as mostclassification algorithms, in its standard form it is only suitablefor numerical input data. Many sensors which are available on currentinformation appliances are nominal or ordinal in type, making theiruse difficult. Additionally, the automatically created clusters areusually too fine-grained to distinguish user-context on an applicationlevel. This paper presents general and heuristic extensions to theLLGNG classifier which allow its direct application for context recognition.On a real-world data set with two months of heterogeneous data fromdifferent sensors, the extended LLGNG classifier compares favorablyto k-means and SOM classifiers.