Retrieving Pre-Embedded Text in Qdrant’s “pageContent” Field

[DongHan_Kim]

I manually loaded a Markdown document into a Qdrant collection with arbitrary metadata, without using LangChain. Although embedding-based search is working, the “pageContent” field is being returned as empty.

Which field name should be used in the payload of a Qdrant Point to ensure that the original text (before embedding) is retrievable in the “pageContent” field?

• n8n version: 1.79.3
• Database : SQLite
• n8n EXECUTIONS_PROCESS setting : own
• Running n8n via : npm
• Operating system: Mac OS 15.3.1(24D70)

[n8n]

It looks like your topic is missing some important information. Could you provide the following if applicable.

• n8n version:
• Database (default: SQLite):
• n8n EXECUTIONS_PROCESS setting (default: own, main):
• Running n8n via (Docker, npm, n8n cloud, desktop app):
• Operating system:

[Nemo]

I have the exact same use case
Did you find a solution ?

Maybe one option could be to import a collection through n8n into Qdrant, And inspect from the quadrant dashboard to compare the structures

I ll give it a try

[Nemo]

Ok, I have some good news for you

I did a import into a “testing_collection” using Qdrant Vector Store - Insert document, using the chat input

I saw the structure in the Qdrant collection

And I tested this Qdrant collection in my other flow (where I had empty pageContent) and I was able to successfully retreive the correct content

Then, in my TS file (from which I manage my own imports into Qdrant) I updated my script with

let point = {
        id: id++,
        vector: "my array of vectors produced by embeding model",
        payload: {
          content: "my initial string content" ,
        },
      };

And now it works

[DongHan_Kim]

Holy. It resolved.

Thx a lot!!