Data gained through eye-tacking experiments is commonly delivered as CSV file. In order to store, manipulate and update this data in a single data model, we propose the usage of a general graph schema for all kind of eye-tracking data. This not only enables extending and updating the data but also makes cooperative work possible. As eye-tracking data resembles highly interconnected data, the usage of a graph database is beneficial. We propose a general, reusable graph schema to adequately handle eye-tracking data for any use case. This schema consists of two levels. A metadata level holding additional data about test persons (age, profession...) and a time series level with the eye-tracking data. To prepare the data for ML-based analysis, we implemented an in-graph-database data preprocessing pipeline with a human-in-the-loop approach. For each preprocessing step, at least two operators are available that can be chosen depending on the data and the use case. All code snippets are implemented with Neo4j's query language Cypher.