The web is full of data sources that we wish to manipulate on a large scale. The current approach is to represent this data in the form of a data or knowledge graph; for example, open and connected data (open data), social networks, online encyclopedias. This approach is even present in the major Web industries, Alphabet (in Google) and Meta (in Facebook).
The benefit of knowledge graphs is that they can be questioned using logic languages but also structural properties can be learned from them.
Even if knowledge graphs are very important tools for the management of web data, not all data on the Web are edited in such a model. It is therefore necessary to search and learn from text and other less structured content to build new graphs.
This course introduces the main steps a data science engineer needs to take to extract knowledge from large volumes of data.
It will familiarize you with concrete tools for:
Handling and visualizing graphs.
Classify nodes and subgraphs using graph embeddings.
Reasoning in knowledge graphs, using Semantic Web technologies.
Weaving connection graphs between texts and concepts, using semantics.
Mining textual data.
The first 6 sessions will be dedicated to the presentation of concepts and tools, then you will realize projects in pairs.
- Teaching coordinator: Balalau Oana