I am having a csv like (a sample created to state my data)
Thanks in advance.
Post the MERGE cypher you're using to create this.
A common slight error with merge is that it merges the full unique pattern. So if you're merging longer paths, you can end up with duplicates if sub-paths (like 12 to 7 here) occur in multiple different longer paths.
A solution is to break down your load into a set of pairwise relationships; think about always merging
(a)-[:r]->(b) and don't write merges like
(:a)-[:r]->(b)-[:r]->(c) because it could duplicate the a-b relationship