If you'd like to use the site and help me test it, you can join here.

Friday, October 31, 2014

What Makes or Breaks a Sewing Pattern Database Entry

So... in my last post I mostly just said what is going on and that it's a massive task (building a huge database with information about sewing patterns). But I didn't really say how and why exactly we're doing it from a bird's-eye point of view. And I actually wanted to say something about how a good pattern entry looks like.


  1. Have you ever had difficulty finding a pattern for a project idea you had?
  2. Do you own a lot of sewing magazines and lose track of what patterns are in what magazine?
  3. Is your stash so large that you could use a site to help you organize it?
We're building a site where a lot of interesting sewing information comes together.

Patterns are linked to the sources (books, magazines) that contain them. Patterns are linked to their size charts so that you will be able to search for all patterns that use a certain size chart and are available in a certain size. We're indexing the data in such a way that you can really drill down into the search results to find exactly what you're looking for.

The center of the site are your stash your projects. But it doesn't end there. You can link fabrics and patterns (or even pattern views) - allowing you to move along these links to get to the related information. Add a magazine from the pattern sources database to your virtual stash and the site lets you browse all the patterns that are contained in your stashed magazines!

The magic lies in that the site understands dressmaking enough that it can give you this powerful search. A search that lets you actually find the things you're looking for - without having to click through a huge load of useless results (as found on Google when you're searching for patterns).

We're building something that you can truly make your own by setting up your own space for yourself and your friends within the site.


We're starting with a smallish set of tags (currently around 250) that can be attached to projects and patterns. These tags talk about sleeve length, neckline style, waistline style, and so on and so on. It's detailed enough to be able to do interesting pattern/project searches, e.g. for "ankle-length dresses with v-neckline and empire waist".

In addition to the tags recognized by the site, we can tag projects and patterns with any keywords we think suitable - and when we recognize a popular tag, we'll make sure to gather it (and all of its synonyms) into a site-wide category. The system is flexible enough to be able to handle that.

We're adding brands, publishers, magazines and books to the database and linking them up to the patterns. That lets us search for patterns by brand, by magazine, by book, by publisher. It's really quite simple.

So what does a pattern entry look like?

A pattern entry on the site currently consists of the following:
  1. A name. For commercial patterns sold in envelopes, there's often a pattern code - in this case, the name starts with the pattern code and proceeds with a short description of what the pattern is for, e.g. "2498 Dress".
  2. If applicable, the brand under which the pattern is published. The system automatically puts the brand name of the pattern in front of its name whenever the pattern gets shown.
  3. If applicable, the size chart of the pattern.
  4. If applicable, the designer(s) who designed the pattern.
  5. If applicable, the source(s) in which the actual pattern can be found (usually a magazine or book).
  6. If applicable, the fabric requirements of the pattern as specified on the pattern.
  7. The official URL of the pattern.
  8. A list of pattern properties. Pattern properties are things like whether it's a "paper pattern", a "print-and-tape PDF", or some "pattern drafting instructions", whether it's designed "for woven fabrics" or "for knit fabrics".
  9. A list of finished object attributes. These are the tags that describe the object that can be made from the pattern, i.e. here we have the tags mentioned in the previous section
  10. Optionally, a pattern can have several pattern views that each describe the different variants of the pattern specifically. While the finished object attributes recorded on the pattern apply to all the views, it's possible to add additional finished object attributes to the individual views. Having a system this flexible is particularly useful for patterns that have completely different views (e.g. pants and top).
  11. Optionally, a short description of the finished object(s).
  12. Optionally, additional notes about the pattern. Currently, we record fabric recommendations as well as notions requirements here.
At some point, we'll extend the system so it lets us record fabric recommendations and the list of notions needed in a searchable way.

What makes a good pattern/source entry?

A good entry...
  1.  has the relevant tags - so it can be found through the pattern search.
  2.  is named in such a way that it's easy to put it into your virtual stash. That is, it's named exactly like written on the envelope / magazine (wherever possible). When you enter something to your stash, you get a dropdown menu that lets you search through the names of the patterns.
  3. links to the specific brand that is displayed on the pattern envelope / magazine. I.e. when I'm adding a "Vogue Easy Options Custom Fit" pattern, I link it to the brand "Vogue Easy Options Custom Fit" - a sub-brand of Vogue Patterns.
  4. has projects attached. With your linked project, you're helping others to find out whether a given pattern is for them or not. You're also helping others with your notes on whether the pattern instructions are clear and whether there were any troubles with the pattern or not.
In short, a good pattern entry is useful. Anything that makes a pattern entry more useful makes it better. :)

Monday, October 27, 2014

Starting to Enter Patterns to the Pattern Database

The last weeks I have been busy figuring out how to make a halfway-decent faceted search. Making sure I can provide a real search for patterns was the most important thing for me - a database that isn't searchable won't realize its full potential. I've read up on a lot of technologies and libraries and found a setup that works. And I made a very simple pattern search.

Now that I know I can deliver the pattern search, the larger-than-life challenge begins: Filling the database with data. It's something I can't ever hope to achieve alone. There are some obvious reasons and some less obvious reasons why this task is so monumental.

There's no way I can record all the patterns

Let's start with a pretty straightforward point: When I look at Ravelry's pattern database today, I see almost 500,000 patterns. I think it is pretty safe to assume that the sewing world isn't much smaller than the knitting/crocheting world. Even if I spent only 5 minutes entering a pattern, I'd be busy for more than four years assuming that I'd need only 8 hours a day for sleeping/eating/showering and that I had all the patterns available to me.

There's no way I can hunt down every single pattern

Okay, so even assuming I somehow could enter things a lot faster the assumption that I have access to all these patterns is flawed: There are so many patterns that aren't even available anymore that I'd be hard-pressed to find them at all (my pattern stash is tiny). All these interesting patterns are out there. So let's look at what different types of patterns we have regarding availability:
  1. Patterns that are still available and can be found online.
  2. Patterns that are available but can't be found reliably online (they can still be ordered through a catalogue or bought at a store or obtained through eBay / Etsy / etc).
  3. Old patterns that are almost impossible to find (e.g. patterns for authentic historical garments).
There's really no way I can collect them all. :)

Even for the patterns that can be found online, data is lacking

So, case 1 (the pattern is available and can be found online) is a pretty good situation. Most pattern companies do have websites and you can find data about patterns on them. However, the quality of the (textual) data is often lacking. In an ideal world, I would find all the pattern data on the web page of the individual pattern in a way that I can simply import this and make it searchable. However, there are many official pattern pages that simply show images or have only very very little textual description of the pattern. Apparently, the pattern companies are lacking the resources to set up good web pages with all the information we'd want to see there (I'm sure they would make better web pages if they could - after all, they want us to find the (available) patterns.)

Okay, so what's the plan?


Regarding recording and finding patterns:

I suspect you know what I'm getting at here: No single person of us has a chance to record all the patterns - even just recording all those in your own stash can be a huge effort. But the more people come together and share, the less every single one of us needs to do.

Regarding the lacking quality of pattern data:

So.. what kind of data do we actually want for patterns? I think it would be awesome to browse for patterns and projects based on a common set of tags (and later also based on fabrics and fabric recommendations, as well as available size). For that, it helps a lot to
  1. record meaningful, standardized tags for each pattern.
We have a partially-complete set of such standardized tags on the site already - and we update this set as we discover new tags of common interest. Since every tag has a specific meaning, when you search through tags, you get very precise search results.

Why should I consider helping?

Have you ever helped someone who was searching for a pattern with specific properties for a project? This is similar: by recording patterns in a freely-accessible database, you're helping people find patterns in a highly scalable way - next year at the same time, when someone asks where to find a pattern you can point them to the database.

Do you have a huge pattern stash and need to get it organized? You have huge stacks of magazines and they pile up in your sewing space - yet at the same time, when you're looking for a pattern for your next project, you have trouble finding what you're looking for? On the site, you can record all your magazines, books and individual patterns into your "stash". In turn, the site gives you a search page that lets you search through all the patterns that are in your magazines and your books (as well as those you own individually).

Things are moving

We've just begun entering patterns. There is a lot to do in terms of building the functionality and the design of the site. There are a lot of great ideas what could be done.

Nowadays, (when I'm not blogging or making an account for someone) I'm constantly working on improving the pattern entering process and making things more usable. I'm very thankful for having people with me who can point exactly at the things that need to be improved.

Do you have a huge pattern stash? Are you a librarian in spirit? We could use your help if you can spare the time. :)