Tell me what’s been done before

In GMail, Google has an amazing knack for advertising products that are relevant to the email that you’re writing. They have a system that analyses the language of what you’re typing and looks up related information in their database.

What if we apply this idea to a users own content, so that work they have previously produced is ‘advertised’ to them as they write their documents, forum posts, blog entires.

Here’s a couple of examples:

Example 1: Creating a document in a workgroup

Person A starts to write a work plan for a new project. They start typing in the heading for the document ‘Work plan for the new jumper museum in Nottingham’.

At this point, the computer starts to look through content (documents, emails) that Person A has already created, and can look to content created by other members of Person A’s team.┬áThe computer ‘suggests’ these documents to the user in a side window, in this case it shows “Person B wrote document ‘Proposal for the new jumper museum in Nottingham’ 1 month ago”. In this situation, obviously we want to look to that document to get some ideas of what needs to be done.

This idea can then be extended to individual sections of the document, e.g.

  • Writing a power plan could suggest previous documents that contained power plans
  • Writing a biography about Person B could suggest previous times you wrote a biography about that.

The main idea is trying to remember items that the user has made before, and suggesting to the user that they go back and look at those things when they’re trying to do something similar. This is based on the premise that eventually, the ser might forget about the previous work, or it may have been carried out by someone else in the team without their knowledge.

Example 2: Forum post

I’m in a forum asking a question. As I start to type my question, i get suggestions of posts that have already been made that relate to that question. Perhaps somebody’s already answered my question and by wording it out myself, the computer can identify which posts are relevant to my query.

Example 3: Academic documents

I’m writing a document about, let’s say using OpenCL to do soft body mechanics. As I write about the different bits of the document, the suggest box can tell me about academic papers related to the specific topic that i’m writing about. In this case, it could also act as a portal to transact those papers.

So what’s the difference between this and just searching before you start writing? Well, not much. But it does have 2 advantages

  1. The search is fit automatically into your workflow, so it does it even if you forgot about it, and it does it for all the little bits that you’re making up as you go along
  2. The fidelity of the search is increased as you are inputting more data that the search can be based on

Leave a Reply