Difference between revisions of "Talk:SiGNaL³"
| Line 2: | Line 2: | ||
As an example we will take a tour of Wikidata starting in alignment with the original [[SiGNaL]] article of 2018. | As an example we will take a tour of Wikidata starting in alignment with the original [[SiGNaL]] article of 2018. | ||
| − | [https://www.wikidata.org/wiki/Q2 Earth (Q2)] and [https://www.wikidata.org/wiki/Q405 Moon (Q405)] | + | [https://www.wikidata.org/wiki/Q2 Earth (Q2)] and [https://www.wikidata.org/wiki/Q405 Moon (Q405)] have a relation that can be described as an edge |
| + | in a knowledge graph. When visiting the node Earth of the knowledgraph an appropriate visualization of the node and relevant possible paths/edges should be | ||
| + | visualized. With the power of state of the art [https://en.wikipedia.org/wiki/Large_language_model Large Language Models] the selection of the visualization | ||
| + | should be feasible in a way that is superior to traditional knowledge graph exploring approaches. | ||
| + | |||
| + | The [https://www.wikidata.org/wiki/Q843864 Earthrise (Q843864)] picture is a symbol for the relation between Moon Earth and [https://www.wikidata.org/wiki/Q5 Human (Q5)]. When mankind was capable of taking this viewpoint it was a groundbreaking moment in human history and the picture is iconic in the Semiotic sense. | ||
| + | |||
| + | Triples of two nodes linked by a relation are the core elements of knowledge graphs (KGs). Navigating such a graph in a human friendly / natural / simple way is mandatory for successful usage of such knowledge graphs. | ||
| + | |||
| + | The strategic move now is to make sure relevant subgraphs are readily available to fullfill needs of projects based on the usecases and usecase scenarios of a project. Such scenarios fortunately can be describes with a sets of situation/action/expected result descriptions. A situation describes a relevant subgraph in a concrete state. The action describes what a system should be offering as an API working on the situation subgraph. The expected result is describing the result subgraph potentially with modifications that have been applied. | ||
| + | |||
| + | Natural language usecase descriptions may be transformed to augmented knowledg graph traversals this way. The journey thru the knowledgraph may be interactive. | ||
| + | A first application is knowledge graph exploration. The traversal path and the human input gathered on the way may now be used to e.g. create a relevant subgraph and what is more important an abstract description of the subgraph that allows to create a local copy that the project stakeholders may work with e.g. by modifying and extending. Synchronization of this local KG with the original KG (respecting privacy settings ...) is useful. | ||
Latest revision as of 06:34, 10 June 2025
SiGNaL³ will apply Retrieval Augmentation Generation cleverly. As an example we will take a tour of Wikidata starting in alignment with the original SiGNaL article of 2018.
Earth (Q2) and Moon (Q405) have a relation that can be described as an edge in a knowledge graph. When visiting the node Earth of the knowledgraph an appropriate visualization of the node and relevant possible paths/edges should be visualized. With the power of state of the art Large Language Models the selection of the visualization should be feasible in a way that is superior to traditional knowledge graph exploring approaches.
The Earthrise (Q843864) picture is a symbol for the relation between Moon Earth and Human (Q5). When mankind was capable of taking this viewpoint it was a groundbreaking moment in human history and the picture is iconic in the Semiotic sense.
Triples of two nodes linked by a relation are the core elements of knowledge graphs (KGs). Navigating such a graph in a human friendly / natural / simple way is mandatory for successful usage of such knowledge graphs.
The strategic move now is to make sure relevant subgraphs are readily available to fullfill needs of projects based on the usecases and usecase scenarios of a project. Such scenarios fortunately can be describes with a sets of situation/action/expected result descriptions. A situation describes a relevant subgraph in a concrete state. The action describes what a system should be offering as an API working on the situation subgraph. The expected result is describing the result subgraph potentially with modifications that have been applied.
Natural language usecase descriptions may be transformed to augmented knowledg graph traversals this way. The journey thru the knowledgraph may be interactive. A first application is knowledge graph exploration. The traversal path and the human input gathered on the way may now be used to e.g. create a relevant subgraph and what is more important an abstract description of the subgraph that allows to create a local copy that the project stakeholders may work with e.g. by modifying and extending. Synchronization of this local KG with the original KG (respecting privacy settings ...) is useful.