One More Input to the Nonprofit Planning System

I said in the prior post that two knowledge artifacts are required to run the Grant Generator: the Business Plan Tree, which provides the content, and the Foundation Sequencing Table, which provides the assembly instructions. There is one other input, a community asset that I build and maintain along with the Sequencing Table, which I call the Nonprofit Element Attribute table.

In the process of building the Nonprofit Planner and Grant Generator, there were certain design and formatting decisions that were left to me as the knowledge author. One was the rules for when to use the green 'completeness' highlighting in the business plan, and the other was how to format the output in the grant application.

First, I'll explain the completeness rules. Remember, in KnowtPlan, when an item is completed it is outlined in green in the note view and highlighted in green in the outline view.

The rule for when to apply the highlighting to the 'leaves' of the tree is simple: when both the short and long answer are filled in, add the green outline. So the completeness rule for 'Organization's Mission Statement' is 'self'; it is highlighted as soon as its 'self' is complete.

The rules for how to roll the highlighting up to higher levels in the tree are somewhat more complicated. That is because some of the higher level notes require their own answers and some are just headers added for organization purposes.

An example of a note that is just a header is 'SWOT Analysis'. It has no content; if you clicked on it, you would not get a popup window asking for a long and short answer. So its completeness rule is 'children'; it is highlighted as soon as all of its children are highlighted. In the screenshot above, SWOT is not highlighted because two of its attached notes have not been filled in yet.

On the other hand, 'Organization's Purpose' is a note that acts as a header but also requires its own content. The charity is expected to fill in the organization's purpose, which should be short and compelling - different from a mission statement, which can be rather wordy. So the completeness rule for 'Organization's Purpose' is 'both': it is considered complete when all of its children are completed AND its own content is filled in.

So one piece of information I have to provide is the completeness rule for every item in the plan. I do this by adding an attribute called 'completeness' and setting its value to 'self', 'children' or 'both'. A second attribute I need to provide for each item is how it should be formatted in the grant application that is generated. This is similar to assigning a paragraph style in Microsoft Word.

Even though we ask for a 'long' and 'short' answer for each question, it is clear from working with the actual content that some answers tend to be much longer than others. For example, contact info tends to be very short. Descriptions of the Board members, on the other hand, tend to be much longer.

So we created two different paragraph styles: 'inline' puts the question header on the same line as the response pulled from the business plan; 'paragraph' puts the header on a separate line. There are also headers, the ones that have no content of their own, that never show up in the generated grant application. Their style is set to 'none'. This became a second attribute.

I'm sure there will be other attributes that will surface as we continue to develop the Nonprofit system. To make it easy to expand and maintain this sort of information, we capture it in a table that is easy for me to change, and doesn't require changes in computer code. The current version of the Nonprofit Element Attribute Table looks like this:

The Grant Application Generator

While the process of creating a 'composer' that generates grant applications highlighted the need for charities to do better planning, I have never lost sight of my original intention. Yes, the Nonprofit Planner has value on its own, but the magic of the tool is that it collects important content in a highly reusable form. The first composer we have built to utilize this content is the Grant Generator, and there will be other composers down the road that can use the planning content as well.

In my book, and in prior posts about the Recommender, I talked about knowledge patterns. I have defined a knowledge pattern as a consistent set of knowledge artifacts that can be used over and over again to solve a certain type of problem. The knowledge pattern for a Recommender looks like this:

I've written a series of posts about applying this pattern to build a Whiskey Recommender.

The pattern for a Composer is remarkably similar to that of a Recommender. The pattern for a generic Composer looks like this:

How this general pattern translates into specifics for the Grant Composer is shown here:

Two knowledge artifacts are required for a Composer: a document content tree, which contains the raw material for the document to be composed, and a table, which contains the assembly isntructions. 

In earlier posts, I described the Foundation Sequencing Table, which holds the assembly information necessary to generate a grant application. This table will be a community asset that I will build and maintain. There will be only one version of this table because the information requirements for a particular foundation are the same for all charities that apply for grants.

The content to be assembled, however, is unique to each charity; each will have its own business plan, documented using our planning tool. This plan is an asset that will be built and maintained by the charity.

These two knowtifacts, then, are inputs to the composer application. Like all of our applications, the first step in using the tool is to collect the context for the current situation. In general, we run an application to solve a particular problem in the current moment, and the context is the way we describe our current needs. 

The Grant Application Generator needs to collect two types of context information.

The first type is information that helps the application 'prune' down the input knowledge artifacts to the relevant columns, tree branches and fields. Questions like 'which foundation', 'which program' and 'long or short answers' fall into this category.

The second type is specific information that refers to this particular grant request only, and therefore cannot come from the business plan. Information in this category includes the request amount and date. This information must be collected at run-time so it can be filled in at the appropriate locations in the generated document.

Once the context is collected and 'OK' is clicked, the Grant Generator creates a document that meets the specifications: just the information the target foundation wants to see, and in the order they want to see it. (I've grayed out this charity's info, to protect their privacy.)

In my next post I'll discuss some of the details about how this report is formatted, and how charities can use it.

Lessons from the Field: My Initial Experience with the Grant Proposal Content Table

One month ago I began using Microsoft Excel to capture the answers of several small charities in the Content Table that provides the 'raw material' for the Grant Proposal Generator. This Content Table was described in my October 7th post.

As is always the case, the real-life application of tools and ideas surfaces many opportunities for refinement. That is, after all, why we do testing! In the last month I became more convinced than ever of the need for such a tool, and the value it can provide charities. But I learned a couple of important things - one about the need to be filled, and another about execution - that will impact our approach.

First, the need. From the beginning, I approached this as an opportunity to help charities generate grant proposals quickly and easily. We would do this by capturing their answers in a format that could be automatically configured to meet a foundation's requirements. What I didn't anticipate is that the charities would also need help developing those answers. In fact, by removing the impediments to proposal construction, we shone a light more brightly on the quality of the answers that were being assembled.

In other words, by focusing our efforts on capturing the one, best answer to a set of comprehensive, rigorous questions it became clear that those answers often did not exist. Or many versions of the answers existed, in many different documents. What became clear to me is that the first need the charity had was support for the business planning process. Without that, our automation efforts would churn out a high volume of low quality grant proposals.

The second thing I learned is that Excel was not the right tool for this job, primarily because a table is not the best shape for the knowledge we are collecting. The content required is a business plan, and the business plan is best described as a hierarchical model, not a table; it is a triangle, not a square. Here are some of the signs that told me a change in approach was necessary.

First, I love Excel, but it is designed to work with numbers not large blocks of text. Trying to type long descriptions into spreadsheet cells was pure misery! That's okay; I had a more fundamental reason that Excel was not the right tool for the job.

Just as a reminder, this is what the structure of the Content Table looks like:

The table has about 150 rows/questions and two columns for General Operating Support (long and short form), and two columns for each program described in the plan. So, if a charity has two programs they will have a total of six columns.

I said up front that the answers to some of the questions will be the same for GOS and all programs - contact information falls into this category. The mission statement of the organization is another example of an answer that doesn't vary. Multiple columns were required because many other important questions will have different answers for different programs: questions regarding populations served, communications strategy, and volunteer requirements, for example.

It's not until you are actually copying and pasting the same answers over and over that you stop and say 'wait! there must be a better way'. And there is.

The copying and pasting was a sign to me that I was using the wrong shape to capture the necessary knowledge. This was not a list of questions (a 'line') with multiple variations (a matrix or 'square'). It was a hierarchical business plan - a tree or 'triangle'. Some of that information I was copying and pasting many times was high level information that only needed to exist once for the whole organization, and sometimes it was information from a different branch of the tree and didn't need to be duplicated for each 'program' branch of the tree.

I knew I needed to rethink the structure of the content knowtifact. I had picked Excel because it was convenient and familiar, but our own KnowtShare application, which makes it easy to construct hierarchical models stored in JSON (a web-oriented data interchange format) was a better choice. So I went back to the drawing board to reinvent the content knowtifact for the tool we are now calling the Nonprofit Planner. My next blog posts will be about that reinvention process.

The Sequencing Table for the Grant Proposal Generator

The Content Table described in my last entry is the 'parts catalog' in the document configurator I want to build. The Sequencing Table will act as the assembly instructions.

The format of the Sequencing Table is similar to that of the content table, because they share a key field: the grant application questions.

Every row in the Content Table must have a corresponding row in the Sequencing Table for that content to be included in the grant proposal that is generated. There are a few headers added for organization purposes, like the "General Questions" text shown here, but they do not show up in the generated proposals.

A number in the column simply means that a given item is in that position in the proposal for that foundation. A blank means the information item is not required. In this way, the column of numbers shows both the subset of information required and the order it should be placed in to meet that foundation's needs.

This column of information needs to be filled out for each foundation that appears in the Grant Proposal Generator. My strategy for capturing this information is to prioritize foundations based on the needs of the charities that participate in our pilot program. I am working closely with the pilot program charities to help them find like-minded foundations that will have a higher probability of granting their requests. I will use the timelines generated from these prospecting lists to prioritize the foundations and schedule their addition to the system.

My goal is, eventually, to have foundations add themselves to the system by providing an intuitive, easy-to-use tool that helps them select the questions they want and arrange them in their preferred order. They will also be able to use the tool to modify their 'questionnaire' at any time. This will be the most efficient way to develop and maintain the Sequencing table, and will guarantee the Grant Proposal Generator is up-to-date with foundations' current requirements.

My next step is to work with my pilot charities to fill out the Content Table.

Creating the Content Table for the Grant Proposal Generator

My first challenge was to identify the comprehensive set of information items that would cover all the questions asked by foundations in their grant application forms. I started with the most comprehensive single form I could find: the one used by the Catalogue of Philanthropy, a Washington D.C. organization, to vet charities in the D.C. region for inclusion in their publication.

I decided based on my initial assessment that a table was the right shape for this information, and that I would use Excel to create my table.

I used the questions as the row labels, and created columns for the various types of proposals a charity would typically generate.

When a charity makes a grant request they must declare how the money will be used. Is it for general operating support? Or is is for a specific program or project? If it is for a specific program, which one? Once the intended use is declared, all of the following questions must be answered with that end use in mind. Some of the answers, like contact information, will be the same for both general support and specific programs, but many will be very different. To accommodate this variation, I setup separate columns to represent the various versions of the proposals. A charity will fill out a project-specific column for each project that requires funding.

I also setup separate columns for long and short form answers. This is necessary because many online application forms limit the number of words for each response. The approach I envisioned here is that the long form would be written first - it's easier to be wordy - and then a short, pithy version crafted next.

I started with the Catalogue for Philanthropy questions, which are in white. The next application form I reviewed was the Washington Regional Association of Grantmakers "Common Application", and the incremental questions from that were color-coded in yellow. I proceeded in this way, selecting application forms from large foundations, adding any new questions I found and color-coding them to show the source. As I had hoped, after only a handful of forms, the number of new questions dwindled to one or two per foundation. This consistency is important, because it confirms that an automated approach is truly possible! After reviewing five sources, the total number of information items was 130, with about 50 related to financial numbers.

At the same time I was creating the Content Table I was also working on the Sequencing Table, the other knowtifact required for the Grant Composer to work. I'll describe the Sequencing Table in my next post.

Knowtifacts Required for a Composer Application

As I said in my last post, a Composer is basically a document configurator. Two required inputs for any configurator are a parts catalog and instructions for how to select and assemble the parts to meet the current need. To build a Grant Composer, I needed to design those two knowtifacts  -- first, generic versions capable of generating any type of document and second, specific versions suitable for generating grant proposals.

My high level goal for The Shape of Knowledge -- the book, this blog and the Apprentice Systems website -- is to create an environment where people with knowledge can generate useful applications that embody that knowledge without writing computer code. To achieve this, people must be able to document their knowledge in easy-to-use tools like Microsoft Excel. A spreadsheet is perfect for capturing 'square' knowledge like tables. We created our own tool, KnowtShare, for documenting hierarchical triangle-shaped knowledge like decision trees. Between KnowtShare and Excel (Excel can also capture lines/lists and cubes/3D tables) we've been able to document many forms of important knowledge. The knowledge necessary to create a Composer is two tables, so only Excel will be required.

Then our job at Apprentice Systems is to create the Application Generators - - "apperators" - that can consume this documented knowledge and generate simple but useful applications. This method only works because underpinning the whole process is a highly reusable, generally applicable knowledge pattern that provides discipline and acts as a link between the documented knowledge and the app generator. Yes, two tables are all that's required, but the two tables must meet a consistent set of requirements for the Composer Apperator to work. My task is to design the two tables and identify those requirements.

The first table is the Content Table, and it will act as the 'parts catalog' for my configurator. It will be the comprehensive set of questions and answers that the Composer will draw from to generate a grant proposal. There will be multiple columns in this table, representing the various versions of the content.

The second table is the Sequencing Table, and it will act as the instructions for how to select and assemble items from the Content Table. There will also be multiple columns in this table, providing different sequencing instructions for different potential target audiences.

"To meet the current need" -- establishing the context for a particular document -- will determine which columns to use from each of the two tables. When running the Composer the user will select the version of the content and the target audience for that particular document, and the Composer will generate an HTML document that meets those specifications.

This is the general game plan. The next posts will show how this actually works for grant proposal generation.

The Composer: a Document Configurator

I said in my last post that one of the challenges a nonprofit faces is that every foundation has their own unique grant application. That's not exactly true.

There's at least one group, the Washington Regional Association of Grantmakers (WRAG), that has attempted to create a common application for their members. It is not unlike universities, who have a 'common app' that many colleges use for vetting students. But if you've applied to colleges you know that 1) many schools don't accept the common app and 2) those who do often have a supplement that asks the additional questions each university still wants answered.

This is because different institutions have different wants and needs. I don't blame them for wanting answers to all their questions; they have the right to know what they want to know before admitting a student to their school. Foundations definitely have the right to know what they want to know before giving another organization their money! In the case of WRAG, out of 100+ members only 18 accept the common grant application.

If you look more closely at the grant application process, you will see that while each application is unique, the questions asked are not. After all, there are only so many items of information you can request from a nonprofit organization. Maybe you can ask a prospective student a wild and innovative essay question to surface his or her true character, but most of the answers a foundation is seeking could be pulled from the nonprofit's business plan, if only they had one.

I felt instinctively that if I could develop a comprehensive set of questions, what would distinguish any particular application would be the subset of information being requested and the order in which the questions were asked. In other words, what I was trying to solve was a classic configuration problem.

Apprentice Systems has solved many configuration problems for our clients. Configuration is a design process where solutions are created by selecting from a set of predefined parts and assembling the parts to meet the needs of a particular situation. We've built systems to automate product configuration (part assembly), process configuration (task assembly) and decision configuration (criteria assembly). What was needed here was a special case of product configuration: document generation. We know how to solve that problem by writing custom code. Could I simplify it into a knowledge pattern that would put automated grant writing within reach of nonprofits everywhere?

After six weeks of research, I feel we are well on our way to making this goal a reality. The knowledge pattern that makes it possible I've dubbed "The Composer". The knowtifacts necessary to generate a Composer are two squares: a Content Table and a Sequencing Table. My next posts will describe my design and creation of these knowtifacts, and the challenges and opportunities for the philanthropic community that have surfaced along the way.

Creating the Whiskey Context Decision Tree - Part Two

My last post described the general challenges of building a context decision tree, and the specific challenges surrounding my search for expertise regarding whiskey selection. In the end, I decided I would have to rely on my own ideas, developed during the many hours of research I conducted while building the other inputs to the Whiskey Recommender. My immediate need was to create a context decision tree so the Recommender Apperator could generate a Whiskey Recommender app. Once that example is published, I hope to find some real, live whiskey experts that can help me build other, alternative Whiskey Recommenders.

I created my decision tree using our free KnowtShare web application. KnowtShare was designed to make it easy to build tree-shaped models. We, at Apprentice Systems, know from our many years of building intelligent systems that trees - hierarchical, non-cyclical models - are essential for capturing several modes of reasoning. Classification, composition and multi-pronged evaluation are all best described using trees. The branching behavior of a decision tree is also easy to document with this sort of model.

While some trees are best built bottom-up, a decision tree is usually built top-down. In other words, you decide the first question you are going to ask, list the possible choices, then determine what choices will be presented based on that first choice, and so on. The tree will document all the possible paths through the context collection process, but users will only travel one of those paths each time they use the application. At the end of each path, a 'leaf' on the tree, will be a recommended option - in this case, a type of whiskey.

I decided, pretty easily in fact, what the first question in my decision tree would be: "Are you going to mix your whiskey or drink it straight up?" I think this is a good first question because most whiskey connoisseurs agree that if you are going to mix your whiskey with Coke, it really doesn't much matter what you drink! Just don't waste a good Scotch! Of course, there are those who will mix an expensive whiskey with Coke, and I need to allow for that possibility as well.

Here is the 'mix it' side of my whiskey decision tree:

The next set of choices I present are using a flavorless mixer (water or club soda) versus a flavored mixer. You can still be a whiskey purist, of sorts, if you only use water or club soda. If your whiskey is to become part of a cocktail, however, it will be difficult to taste the whiskey itself - which is perhaps one of the goals of drinking a cocktail.

I'm pretty comfortable with these first few questions; I think I am on safe ground. Now the going gets much more difficult. I knew the 'flavored mixer' path was taking me towards the cheaper whiskeys, such as Canadian and American blended whiskeys, so I created the next set of choices: do you care more about price or status? If the user selects 'status', I decided to recommend either a Kentucky Straight Bourbon (like Wild Turkey) or a Tennessee Whiskey (like Jack Daniels). Picking 'price' leads to a variety of inexpensive American and Canadian options.

The real puzzler is the next set of options for the non-flavored mixer path. Remember, choices have to be presented in a language that represents the user's perspective (this was covered in the last post). Waxing poetic about the process differences between Scotch, Irish Whiskey and Bourbon won't work here! The target audience of the Whiskey Recommender is someone who is new to the world of whiskey, not an expert. Newbies couldn't care less about the finer points of distilling whiskey, especially since their palates probably can't distinguish the taste differences that result. So I decided to keep things simple, and make the next set of choices "do you prefer American products, or not?"

Finally, at the next level, I reached a point in the tree where I relied on taste profiles to make a distinction. American products were split into 'spicy', which leads to American Rye, and 'sweet', which leads to the bourbons. The non-American choice was split into 'light and smooth' (Irish Blended and Canadian Single Malt) and 'smokey and robust' (Scotch Blended and Single Grain).

This is the point where I need to reiterate an important message from the last post: this is all subjective! Even if I were a whiskey expert, which I'm not, this would still just reflect my opinion. A recommendation, by definition, exists in the world of subjectivity. Otherwise, this would be an 'Answerer', not a Recommender.

The power of the Recommender pattern is that I can create "Deb's Whiskey Recommender" just by building this tree in KnowtShare and combining it with the Whiskey Data Table in the apperator. Then my friend Richard, a real whiskey expert, can generate "Richard's Whiskey Recommender" by creating his own decision tree. He can even reuse my Whiskey Data Table. Better yet, in the future we can crowd source a much more complete data table and make that a community asset we all tap into when we build our own unique Whiskey Recommenders. That's the vision.

Back to my Recommender. Let's take a look at the other major branch of my decision tree, the 'straight up' branch:

I reused the 'do you prefer American products' choice again, and early on in the question sequence, because I needed to account for American Single Malts and I can't do it through a taste profile. American Single Malts are all over the board, taste-wise, because American distillers are using a wide variety of processes to create their single malts. Some are following the Scottish methods, and beating high-end Scotches in blind taste tests. Some are creating their own unique processes that are leading to unique taste profiles. All that binds them together is their country of origin, and that American origin is important to some people, including me. I like our underdog status in the Single Malt arena, and I like to support these distillers when I can.

As for the rest of the 'straight up' branch, it is a taste-driven breakdown between Irish Whiskey and Scotch, and within Scotch, a breakdown into the various regions. While some regions have distinct flavor profiles (like Islay) and some are more diverse (like the Highlands), taste is still the easiest way to direct a Scotch drinker to a particular region.

A real Whiskey Recommender, as opposed to this demo application I am building, could recommend specific Scotch brands, like Laphroaig. To keep things manageable, I have decided to stop my recommendations at a type of whiskey, like Islay Single Malt Scotch. If you've read the other posts about building the Whiskey Recommender, you will know that this is one of many simplifying assumptions I have made during this process.

Now that I've built my Whiskey Context Decision Tree and saved my KnowtShare file, I have all of the knowtifacts necessary to generate my app. The next step will be to enter them into the Recommender Apperator and generate my own custom web app.

Creating the Whiskey Context Decision Tree - Part One

Now that my Whiskey Data Table (or Whiskey Options Table) is complete, I can turn my attention to creating the context decision tree. The context decision tree in a Recommender contains a series of questions the user will be asked to determine his or her needs and gather any information about the current situation that may impact the decision being made. It also contains the link between each path through the decision tree and the options, or rows, in the data table.

Given these two important roles -- determining the user's needs and pointing to a particular set of options or solutions -- it is not surprising that the context decision tree is the knowtifact that contains the most critical 'expertise' in a Recommender application. In a complex subject like whiskey, there are hundreds of questions one might ask before making a recommendation, and even then it is not obvious what the best recommendation should be. In other words, creating a context decision tree is hard and full of subjectivity.

Because it is such a subjective exercise, if you ask ten different domain experts to build a context decision tree you are likely to get ten different results. This diversity of opinion reflects real life; if decisions were simple and obvious, we wouldn't need to consult with experts. The value of capturing expertise in a Recommender is multifaceted: it makes knowledge available to a wider audience by embodying it in a web app, it makes it possible to compare and contrast the decision processes of multiple experts, and multiple experts can leverage the same options table to build their own Recommender just by building different context decision trees.

The single biggest benefit of creating a context decision tree, however, is both subtle and surprising, and that is that it forces experts to actually articulate their knowledge. You might think getting experts to talk about what they know is not such a difficult thing to do, and that is true. But getting them to create a context decision tree can be very tough, and it raises issues like 'tacit vs explicit' knowledge and 'shallow vs deep' knowledge.

First, much of an expert's knowledge is typically tacit, especially when it comes to their decision making process. They just 'know' what to do, and they often can't explain why they do it. Building a context decision tree requires that they make this tacit decision process explicit.

Second, much of what passes for expertise is really 'shallow' knowledge like product and historical information, and rote process knowledge. This sort of expertise only requires a good memory. Building a good context decision tree often requires 'deep' knowledge, knowledge not only about what and how things are but why things are the way they are. It requires the ability to apply knowledge, not just repeat facts.

I've said from the beginning, I am not a whiskey expert. I started down the road on this demo application because I found a really cool classification chart for whiskies. I was able to take the next step and build the Whiskey Data Table because collecting data is more about diligence than expertise. But now I needed to find a whiskey expert.

Once again I turned to the internet, which is chock full of websites that claim to contain whiskey expertise. I started doing searches with phrases like "how to pick a whiskey" and "how to select the right whiskey". The search engines were able to pull up hundreds of links that purported to answer that question, but when I read what they had to say it was always a regurgitation of the same information over and over -- the differences between Irish whiskey and scotch, between Kentucky straight bourbon and Tennessee whiskey, between the bourbon process and the scotch process, between the mash bill for Canadian rye whiskey and American rye... The closest anyone came to providing useful information for the task at hand was the taste profiles for different scotch regions. Yes, taste is something a beginner whiskey drinker would care about. Mash bills? not so much.

Which brings us to a classic marketing problem. Users/consumers/customers are always looking for benefits -- touchy-feely, use-case specific benefits that will accrue to them as users. Manufacturers/sellers/retailers tend to want to talk about product features -- concrete, well-documented facts about the item or service they are selling. The marketing people are the ones who must make the connection between the two, and advertising and product literature exist to create that bridge.

In a similar fashion, a context decision tree, which is a series of questions that will be answered by the user, needs to be written in the language of benefits and use-cases. My web research surfaced very little knowledge that would help someone new to the world of whiskey decide where to begin. With so many choices, some of which are quite expensive, how can a relatively inexperienced drinker make a selection that is a good fit for his or her current needs? That is what I needed to know to create a good Whiskey Context Decision Tree.

After approximately 100 hours of total research time -- between building the classification tree, the data table, and general whiskey research -- I was going to have to use my own creativity and judgment to build the context decision tree. It was sure to have a lot of flaws, but at least I would be able to generate an initial version of a Whiskey Recommender.

Adding Images to the Whiskey Recommender

While images aren't required for a Recommender app, there is a place for a picture in the 'baseball card' template, just as you would expect for any baseball card. Here is an example of the template, as it is used in the Beach Town Recommender:

I want to add images to the Whiskey Recommender, but this is more difficult than it was for beach towns. The challenge arises, once again, because I am working with a much larger set of possible options -- 450 different whiskey brands!

The Beach Town Recommender has about 30 towns in it, so it was relatively easy to search the internet for good medium-sized images of my beach towns and save them to a folder on my computer. I won't be able to do that for 450 whiskeys; it's just too much work. I'll have to find different images for different types of whiskeys, and stop there.

The basic process I used was to (1) locate images using a search engine, and then (2) save those images to a folder using a simple, consistent naming convention. Then I added a "Picture" column to my data table and (3) entered the file name of the image I wanted to display for that particular whiskey.

I started out with a nice, generic picture of whiskey that I could use for my default image. This is the one I selected:

I named this 'whiskey.jpg' and I copied that name into the entire "Picture" column in my data table. I would copy and paste over this name if I found a better/more specific picture.

Now I moved on to my top level classes in my classification tree: American, Scotch, Canadian and Irish. What sort of images could act as generic images for these high level classes?

I learned after doing a bit of image research, which I did in the image section of both Google and Bing, that there were pictures available that showed groups of whiskey bottles, selected to represent different types of whiskeys. Here's an example:

This is the image I decided to use for 'American'. Someone else had already created this for me -- I could see that different bottles had been cut and pasted into the image -- but that's okay, it saved me some time. What I liked about this picture is that it had a selection of Tennessee Whiskey, Kentucky Straight Bourbon and Rye Whiskeys, making it a snapshot of a variety of American types and brands. I found similar images for Canadian, Irish and Scotch. At this point, rather than copying and pasting image names into my data table, I decided to hold off and see what other pictures I could find.

I was able to find similar multi-bottle pictures for several sub-classes: American rye, Kentucky bourbon, Tennessee whiskey, single malt scotch, and blended scotch. I would have liked to have pictures for the other sub-classes as well, like 'American blended' or 'American corn', but I couldn't find anything appropriate. So I used the generic 'American' image for those types, just as I had to settle for only one image for all of the Canadian whiskeys. At this point, I copied the names of the most specific image I had available into the appropriate cells in my "Picture" column.

Now I assessed the situation. I realized that when someone used the Whiskey Recommender, if they were directed to a particular type of whiskey, like single malt scotch, they would see 20+ 'baseball cards' -- all with different information on them but with the same picture, this one:

That's a little too boring, even for a demo! So I knew I needed to push one more level down in the data, if I could.

There were a couple of different options I could pursue. In single malt scotch I had another column in my data table: "Region". I could look for images that went with Speyside, Highlands, Islay, etc. But the other possibility was to look for images that went with the "Company" dimension, images for Diageo, Suntory/Beam, Sazerac, and the like. I decided to look for company-specific images, because that could add visual diversity to all of the whiskey types, not just single malt scotch.

I was able to find images for Beam Irish, Beam American, Forty Creek Canadian, Sazerac Canadian, Diageo Scotch and Pernod Ricard Scotch. Here is the image for Pernod Ricard Scotch:

After adding these images to my image folder, I copied and pasted the image names into the appropriate cells of my data table. I would love to have a picture of every specific whiskey brand, but for this particular demo, I decided to stop at this point. The Whiskey Data Table -- one of the two knowledge artifacts required to create a Recommender app -- was complete.

When I build the Whiskey Recommender using the Recommender Apperator, I will need to upload my folder of images along with my Whiskey Data Table and Context Decision Tree so that the pictures can appear in my web app.

One final comment about finding appropriate images: I always use my search engine's tools to limit the images I am shown to medium-sized files. That is big enough to provide good resolution in my web app, while saving data storage and reducing load times during execution.

Now I need to move on to creating the final 'knowtifact' that is required to generate a Recommender: the Context Decision Tree.

Creating a Whiskey Recommender: The Classification Tree

The whiskey poster I found in an issue of Fast Company (see my last post) showed a complex network of specific whiskey brands and types, but on closer inspection it's clear that it is a single classification tree bound together by class/subclass relationships. There are four major classes of whiskey in this chart: American, Canadian, Irish and Scotch. These classes then break down into sub-classes in a variety of ways, based on the ingredients and process used.

I needed to capture the knowledge from the Whiskey poster in an actionable form, so I used KnowtShare to create a classification tree. This is the American Whiskey branch:

The window in the lower right hand corner is a navigation pane that shows the complete tree.

The KnowtShare app lets me create notes and group them quickly and easily into a hierarchical structure.

The American Whiskey sub-classes are a mix of grain used (wheat, rye, corn) and process (single malt and blended). Bourbon is a unique designation based on both grain requirements (at least 51% corn) and process (charred barrels).

Some of these sub-classes are further broken down based on region and company. The whiskey poster goes on to name specific brands in each of these categories, but with 450 brands listed I opted to stop the KnowtShare tree at this level and handle the brands in another knowtifact: the options data table.

Here are the Canadian and Irish branches of the tree:

The Irish whiskey branch has two new sub-classes: single grain and single pot.

And finally, here is the Scotch whiskey branch:

Scotch is unique because it calls out six different regions for single malt. Each is considered to have its own special taste based on process, but all are made from malted barley.

In KnowtShare, when working with a very large tree such as this, you can use the page itself as the top node of the tree (here it equals "whiskey"). This allows you to arrange the next group of classes in any way that works best visually. The text view will show everything as one comprehensive outline and the .knt file that is saved will combine the four groups into a single hierarchical file.

The next task will be to create the options data table.

Creating a Recommender

I am going to do a series of posts that document the process of creating an app using a knowledge pattern. I am going to use a particular pattern we call a 'Recommender'. The Recommender pattern is shown in both The Shape of Knowledge eBook and in Part 2 of the video series that provides an overview of the eBook's content.

You can view the videos at these links:
Part 1 https://www.youtube.com/watch?v=ui8kjxGmjE0
Part 2 https://www.youtube.com/watch?v=Hma_ho2fHck

The Recommender pattern is made up of one optional knowtifact and two required knowtifacts.

The optional knowledge artifact is a classification tree of the options under consideration. The two required knowtifacts are a context decision tree and a data table that lays out all of the options and their attributes.

In the book I talk about two simple Recommenders: an app that asks a potential knowledge author questions about what they want to achieve and then recommends the best knowledge artifact for the job (the KA Recommender) and a Beach Town Recommender that asks a future traveler a series of questions and then recommends the best beach towns for vacation.

For this example I am going to create a Whiskey Recommender (or "Whisky" Recommender, if you prefer that spelling of the word; I learned in my research that whiskey aficionados feel passionately about this question).

I picked whiskey NOT because I am an expert - I'm not. I picked it because I ran across this great visual in an issue of Fast Company:

This poster, which is very cool looking, appears to be a hopelessly complex constellation of whiskey names and types, but I knew immediately that it was something much more fundamental -- it is a classification tree for whiskeys.

In The Shape of Knowledge, the classification tree is one of my prime examples of the Triangle shape. This whiskey visual doesn't look very much like a triangle, or 'rooted tree', but it is. My first task in building the Whiskey Recommender is to transform this content into a usable form, and I will do that by organizing the basic tree structure shown in the diagram, using KnowtShare.