diff --git a/docs/sti/default-sti-approach.md b/docs/sti/default-sti-approach.md
index aaf1738..3bc44a2 100644
--- a/docs/sti/default-sti-approach.md
+++ b/docs/sti/default-sti-approach.md
@@ -10,6 +10,6 @@ MantisTable UI intagrate a default STI approach: **s-elBat**. If you want to int
 
 **s-elBat**  (the name comes from taBle-s and Semantic Entity Linking to BAtch Table) is a supervised and unsupervised STI approach.It incorporates various strategies to address all STI challenges.
 
-<img src="/mantistable-docs/img/selbat.png" height="100"></img>
+<img src="/mantistable-ui-docs/img/selbat.png" height="100"></img>
 
 This new approach, s-elBat, builds upon and enhances our previous STI approaches ([Cremaschi et. al, 2019](https://ceur-ws.org/Vol-2553/paper3.pdf) - [Cremaschi et. al, 2020](https://ceur-ws.org/Vol-2775/paper8.pdf) - [Avogadro et al., 2021](https://ceur-ws.org/Vol-3103/paper7.pdf)). Based on the experience gained from those solutions and their involvement in [SemTab Challenge](https://www.cs.ox.ac.uk/isg/challenges/sem-tab/), it integrates a feature vector-based entity disambiguation approach. This approach combines heuristic and machine learning techniques to achieve precise disambiguation results.
\ No newline at end of file