Build chatbot for education purporse - nlp

I try to build pretty simple chatbot which gonna be useful for kids as education tool.
As an example:
Bot: Where do you live?
Expected answer: I live in [Chicago].
Bot: Where is [Chicago]?
Expected answer: It's in Illinois.
And so on..
The main purpose on this bot is check does kid understand simple questions.
Firstly, I hard-coded list of expected answers.
But I thinks I can use for it tool like dialogflow?
Is it smart enough to use that powerful tool for this case?

This is a really broad question, but yes, Dialogflow's Natural Language Understanding engine is certainly powerful enough to be trained with some example answers and extrapolate some other similar ones. It is not perfect, but is better than trying to hard-code and match every possible answer.

I'd suggest you try out Kore.ai bots as well for this very purpose. You can find it in here - bots.kore.ai/botsbuilder
I have been putting myself to train on all the available chatbot frameworks that are there in the market today & I find it convenient to work on Kore.ai platform, more or else based on the use case.
The simple conversation flow - assuming would also have a lot of exceptions, default messaging, interruption handling, session time outs, spelling errors in answers and a proper conversational flow - might be a good excercise over the Kore.ai platform.
Give it a try & let me know before I proceed to spread the word. If you have anymore questions specific to chatbot, please do DM me.
Thanks.

Related

Can you create a custom model using GPT-3 to answer questions only about a specific topic?

I'm using GPT-3 to create a chatbot that can answer questions related to a specific topic. Can GPT-3 be trained to detect questions that are irrelevant to the topic and refuse to answer them?
Example: Let's say I want to create a chatbot that can only answer questions about
Javascript. If it is asked to list the seven wonders of the world, it should refuse to answer.
This has been successful for me, you may wish to try it as well.
I want you to act as a javascript guide. You are here to help answer any questions I may have about the language. If I have any questions, you will do your best to provide a helpful response. Please note that if my question is not related to Javascript, you have to write only "Error". Let's get started.

Meta construction capabilities?

I am currently considering Orange as the base for a meta-learning assistant prototype I intend to develop, but before committing myself to a thorough exploration of the documentation and learning about python development (which would both be quite time consuming), I would appreciate some insight regarding the feasibility of such prototype within Orange framework.
The main aim of the prototype I intend to develop is to allow efficient use of data mining and machine learning algorithm by non experts. Concretely, I wish as a first step to be able to give the user a workflow answering his modelling need, that I elicit from his dataset and expression of his need. In order to perform this elicitation, I intend to run a process that implies designing and executing learning workflows on his data.
Is it possible from within the Orange framework (or else from an above "supervising" framework) to automatically define and execute learning workflows ?
Yes, it is.
We have actually experimented with a "recommendation system" that would suggest parts of the workflow to the user. It wasn't useful. Also, there have been various meta-learning projects in the past and I think that the general consensus is --- it doesn't work. ;)
But if you intend to try it, Orange is suitable platform for this.
#hoijui: Orange no longer has any other mailing list or forum, just this one. Developers follow Stack overflow and answer questions there.

Hello world for spoken NLP interaction (like Siri)?

If I have zero experience developing spoken NLP interaction, what is the easiest way for me to make a Hello World, and begin experimenting from there?
Any platforms and programming languages and APIs are relevant here, as long as they give me the most basic equivalent of Hello World AND are flexible enough that I could potentially play around for a few hours in that environment using various public REST APIs and glueing strings/regex together and get some useful demo apps out of it.
Please offer kind suggestions to improve this question if it's not clear enough, since it's probably a topic on a lot of programmers' minds and something which is not yet mainstream.
If you want to play with building blocks, you can use:
either SphinxCMU or the Google's voice recognition API (used by Chrome) to turn the user's voice into text
the Wit API to extract meaning (the user intent/question) from text
and then our own module to build an answer
Start looking on existing projects like this one:
Pi-Voice
read the code and analyze what they are doing. Those are toys anyway since technology behind Siri is quite complex
To get a deep understanding of the technology read the papers from the CALO project which was a Siri base:
Calo Project Website

Effective strategies for studying frameworks/ libraries partially [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
I remember the old effective approach of studying a new framework. It was always the best way to read a good book on the subject, say MFC. When I tried to skip a lot of material to speed up coding it turned out later that it would be quicker to read the whole book first. There was no good ways to study a framework in small parts. Or at least I did not see them then.
The last years a lot of new things happened: improved search results from Google, programming blogs, much more people involved in Internet discussions, a lot of open source frameworks.
Right now when we write software we much often depend on third-party (usually open source) frameworks/ libraries. And a lot of times we need to know only a small amount of their functionality to use them. It's just about finding the simplest way of using a small subset of the library without unnecessary pessimizations.
What do you do to study as less as possible of the framework and still use it effectively?
For example, suppose you need to index a set of documents with Lucene. And you need to highlight search snippets. You don't care about stemmers, storing the index in one file vs. multiple files, fuzzy queries and a lot of other stuff that is going to occupy your brain if you study Lucene in depth.
So what are your strategies, approaches, tricks to save your time?
I will enumerate what I would do, though I feel that my process can be improved.
Search "lucene tutorial", "lucene highlight example" and so on. Try to estimate trust score of unofficial articles ( blog posts ) based on publishing date, the number and the tone of the comments. If there is no a definite answer - collect new search keywords and links on the target.
Search for really quick tutorials/ newbie guides on official site
Estimate how valuable are javadocs for a newbie. (Read Lucene highlight package summary)
Search for simple examples that come with a library, related to what you need. ( Study "src/demo/org/apache/lucene/demo")
Ask about "simple Lucene search highlighting example" in Lucene mail list. You can get no answer or even get a bad reputation if you ask a silly question. And often you don't know whether you question is silly because you have not studied the framework in depth.
Ask it on Stackoverflow or other QA service "could you give me a working example of search keywords highlighting in Lucene". However this question is very specific and can gain no answers or a bad score.
Estimate how easy to get the answer from the framework code if it's open sourced.
What are your study/ search routes? Write them in priority order if possible.
I use a three phase technique for evaluating APIs.
1) Discovery - In this phase I search StackOverflow, CodeProject, Google and Newsgroups with as many different combination of search phrases as possible and add everything that might fit my needs into a huge list.
2) Filter/Sort - For each item I found in my gathering phase I try to find out if it suits my needs. To do this I jump right into the API documentation and make sure it has all of the features I need. The results of this go into a weighted list with the best solutions at the top and all of the cruft filtered out.
3) Prototype - I take the top few contenders and try to do a small implementation hitting all of the important features. Whatever fits the project best here wins. If for some reason an issue comes up with the best choice during implementation, it's possible to fall back on other implementations.
Of course, a huge number of factors go into choosing the best API for the project. Some important ones:
How much will this increase the size of my distribution?
How well does the API fit with the style of my existing code?
Does it have high quality/any documentation?
Is it used by a lot of people?
How active is the community?
How active is the development team?
How responsive is the development team to bug patch requests?
Will the development team accept my patches?
Can I extend it to fit my needs?
How expensive will it be to implement overall?
... And of course many more. It's all very project dependent.
As to saving time, I would say trying to save too much here will just come back to bite you later. The time put into selecting a good library is at least as important as the time spent implementing it. Also, think down the road, in six months would you rather be happily coding or would you rather be arguing with a xenophobic dev team :). Spending a couple of extra days now doing a thorough evaluation of your choices can save a lot of pain later.
The answer to your question depends on where you fall on the continuum of generality/specificity. Do you want to solve an immediate problem? Are you looking to develop a deep understanding of the library? Chances are you’re somewhere between those extremes. Jeff Atwood has a post about how programmers move between these levels, based on their need.
When first getting started, read something on the high-level design of the framework or library (or language, or whatever technology it is), preferably by one of the designers. Try to determine what problems they are trying to address, what the organizing principles behind the design are, and what the central features are. This will form the conceptual framework from which future understanding will hang.
Now jump in to it. Create something. Do not copy and paste somebody's code. Instead, when things don’t work, read the error messages in detail, and the help on those error messages, and figure out why that error occurred. It can be frustrating, when things don’t work, but it forces you to think, and that’s when you learn.
1) Search Google for my task
2) look at examples with a few different libraries, no need to tie myself down to Lucene for example, if I don't know what other options I have.
3) Look at the date of last update on the main page, if it hasn't been updated in 6-months leave (with some exceptions)
4) Search for sample task with library (don't read tutorials yet)
5) Can I understand what's going on without a tutorial? If yes continue if no start back at 1
6) Try to implement the task
7) Watch myself fail
8) Read a tutorial
9) Try to implement the task
10) Watch myself fail and ask on StackOverflow, or mail the authors, post on user group (if friendly looking)
11) If I could get the task done, I'll consider the framework worthy of study and read up the main tutorial for 2 hours (if it doesn't fit in 2 hours I just ignore what's left until I need it)
I have no recipe, in the sense of a set of steps I always follow, that's largely because everything I learn is different. Some things are radically new to me (Dojo for example, I have no fluency in Java script so that's a big task), some just enhancements of previous knowledge (Iknow EJB 2 well, so learning EJB 3 while on the surface is new with all its annotations, its building on concepts.)
My general strategy though is I'd describe as "Spiral and Park". I try to circle the landscape first, understand the general shape, I Park concepts that I don't get just yet, don't let it worry me. Then i go a little deeper into some areas, but again try not to get obsessed with one, Spiralling down into the subject. Hopefully I start to unpark and understand, but also need to park more things.
Initially I want answers to questions such as:
What's it for?
Why would I use this rather than that other thing I already know
What's possible? Any interesting sweet spots. "Eg. ooh look at that nice AJAX-driven update"
I do a great deal of skim reading.
Then I want to do more exploring on the hows. I start to look for gotchas and good advice. (Eg. in java: why is "wibble".equals(var) a useful construct?)
Specific techniques and information sources:
Most important: doing! As early as possible I want to work a tutorial or two. I probably have to get the first circuit of the spiral done, but then I want to touch and experiment.
Overview documents
Product documents
Forums and discussion groups, learning by answering questions is my favourite technique.
if at all possible I try to find gurus. I'm fortunate in having in my immediate colleagues a wealth of knowledge and experience.
Quick-start guides.
A quick look at the API documentation if available.
Reading sample codes.
Messing around YOU HAVE TO MESS AROUND (sorry for the caps).
If it's a small library/API with a small or no community you can always contact the developer himself and ask for help 'cause he'll probably be more than happy to help you; he's happy that one more person is using his API.
Mailing lists are a great resource as long as you do your homework first before asking questions.
Mailing list archives are invaluable for most of the questions I've had on CoreAudio related stuff.
I would never read javadoc. As there often is none. And when there is, most likely it isnt up to date. So one gets confused at the best.
Start with the simplest possible tutorial you find within some minutes.
Often the tutorial will lead you to further sources at the end, so then most of the time one is on a path that goes on and on, deeper and deeper.
It really depends on what the topic is and how much info is on it. Learning by example is a good way to start a topic brand new to you, especially if you're knowledgeable in other similar libraries or languages. You can take a topic you're familiar with, and say "I understand how to implement using X, lets see how it's done using Y".
So what are your strategies, approaches, tricks to save your time?
Well, I search. I generally never ask questions, preferring to research myself. If worse comes to worse I'll read the documentation. In some cases (say, when I was doing some work with SharpSVN) I had to look at the source, specifically the test cases, to get some information about how the API worked.
Generally, I have to be honest, most of my 'study' and 'learning' is by accident.
For example, just a few seconds ago, I discovered how to get the "Recent" folder in C#. I had no idea how to do that before seeing the question, considering it interesting, and then searching.
So for me the real 'trick' is that I hang around on forums, answer questions, and accidentally pick up knowledge. Then when it comes time for me to research something; chances are I know a bit about it, and searching is easier and I can focus on the implementation [typically implementing a test program first] and progressing from there.

How do you find out what users really want? [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
I've read somewhere (I forget the source, sorry - I think the MS Office developer's blog?), that when you do a survey of users asking them about what features they would like to see in your software/website, they will more often than not say that they want every little thing, whereas collected metrics show that in the end, most people don't use 99% of these features. The general message from the blog post was that you shouldn't ask people what they use, you should track it for yourself.
This leads to an unfortunate chicken-and-egg situation when trying to figure out what new feature to add next. Without the feature already in place, I can't measure how much it's actually being used. With finite (and severely stretched) resources, I also can't afford to add all the features and then remove the unused ones.
How do you find out what will be useful to your users? If a survey is the only option, do you have to structure your questions in certain ways (eg: don't show a list of possible features, since that would be leading them on)?
Contrary to popular belief, you don't ask them. Well, you don't listen to them when they tell you what they want. You watch them while they use what they have right now. If they don't have anything, you listen to them enough to give them a prototype, then you watch them use that. How a person actually uses software tells you a lot more than what they actually say they want. Watch what they do to find out what they really need.
Give them options and the have them arrange them in order of importance. As you said, the users are going to want everything, but this will allow you to tell what they want the most.
You tell them. Then both of you know.
(No, your users won't tell you what they want. That's work. If users wanted more work to do, they wouldn't be looking for software to do their work for them.)
An anecdote from a previous life:
We were planning for a new release and wanted to add some new features to the application. We got the users together and brainstormed what things they wanted to see in the system, placing each "feature" on a yellow sticky on a white board. We then grouped similar requests together and eliminated duplicates or near dups.
We then laid each sticky on a table with a cup in front of it. Each user got 10 pennies to "vote" on the features they wanted. They could put as many pennies in each cup as they wanted, up to all their pennies in one cup if they so desired. We then counted the number of pennies in each cup and chose to implement the top 5 vote getters, in order of votes.
It was surprising to see people that were passionate about a feature while brainstorming and categorizing turn around and not vote for that feature (or vote lightly for it).
Of course, a technique like this will only work if you have ready access to your user base (this was for an enterprise system we developed internally).
You ask them.
(No, you do not know what your users want better than they do. Yes, you will get a lot of stupid answers. Avoid multiple-choice surveys and instead opt for reviewing free-form answers. The information you collect will be invaluable.)
Of course — you could always allow your users to vote on which features they like most...
Users know what they don't want better than they know what they want.
We had brought in a team do do an Oracle eBusiness Suite implementation. They took an interesting approach that had worked very well for them in the past. But it was phenomenal in our environment.
We had cultural issues which meant none of the users were going to stick the necks out to say what they wanted. I had history with the users from the past. Trying to get get requirements out of them was like trying to get blood from a stone. But once you went live the bitching would start.
Anyway the implementation team installed Oracle eBusiness Suite straight out of the box. Give the users the basic training. Then about every 4 weeks for the next 6 months they customized the base installation to accommodate the complaints.
I would recommend against showing them options; as you point out, if it's available, then people will want it just for the sake of having it. Often the users are not aware of the extra costs of developing a particular feature, and just want it because you mentioned the possibility of having it.
The other option is to show a list of all the features you could possibly add, and then attach a price to each one, and then ask users, would it be worth $X to have feature Y, or, how much extra would you be willing to pay for feature Y?
Eat your own dog food
Try to use the application that you write yourself as much as possible. Then you will know how you can improve your application.
According to 37 Signals - Getting Real book, you don't do anything, you don't even record what they want, you just delete mails after one read without any action.
When it comes to implement / fix stuff you'll remember the most important things that your users want from top of your head. Obviously this requires a bit user base.
You need to tie features to cost. Everyone wants features, but not every feature is worth paying for. Ask which features are most important, which would your users be willing to pay for? Develop features based on the priorities supplied by users and stop when they aren't willing to pay for any more. Get the product into their hands as quickly as possible so that you can get real feedback on what doesn't work and what needs to be added. When the users have access to real software, you get much better information. This works best when you are developing specifically for a particular customer. If you don't have access to real customers, consider seeding your product with people (can you say, public beta?) free in order to get better feedback.
Users don't know what features they want. You don't know what features they might be offered. "Features" don't mean anything except as they help them accomplish tasks and achieve goals. And that's where you should start, because they will have a very imperfect understanding how they relate.
There is one thing they know, maybe, much better than you do. And that's how to get their jobs done.
As soon as computer/software concepts and terminology start to leak into the discussion between users and designers, you're off the rails.
So many times users will focus their requirements in terms of what's wrong with, or could be improved about, the software they currently use. Over time, even they lose the distinction between their jobs, and the software they use to do their jobs.
It's a very hard, critically important problem for you to solve this.
The only way to know what the users "really" need is to "be" the user.
Its programming kung fu black belt level.
"Be like water making its way through cracks. Do not be assertive, but adjust to the object, and you shall find a way round or through it. If nothing within you stays rigid, outward things will disclose themselves.
Empty your mind, be formless. Shapeless, like water. If you put water into a cup, it becomes the cup. You put water into a bottle and it becomes the bottle. You put it in a teapot it becomes the teapot. Now, water can flow or it can crash. Be water my friend."
When you be the water/customer, you'll now.
I think Bruce Lee would be a good programmer.
Im very serious. This is the way I work. I cant do things I dont understand, so I have to understand before I do things. When I understand, and my costomers know I understand then I can do a good job. Without understanding there will be missunderstandings. You are the only person who know when you have the correct level of understanding, you are also the person who is responsible to get that knowledge.
The Oracle at Delphi
Pros: accuracy is superb
Cons: if you can interpret the messages, which many people fail to do (often seeing what they want to see). Also requires supplication, which can get messy (contrary to popular opinion, your hecatomb need not be 100 of the same type of livestock).
Psychics
Pros: accurate to a point.
Cons: rare. Prone to mental instability, highly vulnerable to eldritch beings, and might attract unwanted attention from them. Also, it takes experience to sort through the mystery that is the human mind to get to desired information. And sometimes you still need to probe subjects while they're actually doing the thing they need help with, since users lie.
Plant a mole
Pros: New gadgets. New Poisons! Plans within plans within plans. Baby's a freak show. You might learn all sorts of fascinating things in addition to the information you need to help the user.
Cons: Expensive. Chances remain that the agent will turn on you, or fail to learn anything you couldn't learn more simply. If discovered, organization will likely turn or liquidate the asset, which represents a huge investment of resources. Organization might reciprocate.
Guess
Pros: Take a group of people with average to great imaginations and problem solving skills, give them some booze and inspire them with some quotes from Ghostbusters, Big Trouble in Little China, or The Big Lewbowski. Who knows where it will go, but it'll be fun and they might produce something interesting/useful.
Cons: Chances of meeting user's needs are higher than you think, but not that good.
Ask the user
Pros: users feel empowered as part of the process.
cons: until they have to decide on anything, at which point you are on your own. Unless the user is a very experienced user, in which case they probably have a good idea of what the want. There's only like 4 experienced users on the planet though, and nobody ever knows anyone who gets to do a job for them. They may be mythical beasts.
Pretend you care and ask the user (even though you don't really), and then observe them doing whatever key workflow/process/etc is involved and pay attention to what they do.
Pros: you trick the users into thinking their opinion matters, which empowers them but doesn't deliver any other baggage. Since users lie - no purposefully or maliciously mind - you actually get to see them in action and get a better grasp of what the problem is, thus giving you a better foundation for building a solution. Also, you avoid the psychic route, and thus avoid a long and winding road that begins with promise but ends with you and the psychic being eaten by some monstrous, unspeakable thing that is not of this world. Observing the process is like totally Zen, which is good for your Developer Mystique.
Cons: No road trip to the Oracle (which would be EPIC). Spies are much sexier; chicks dig spies. Ghostbusters|Big Trouble in Little China|The Big Lewboski probably aren't involved. Feels more like work than the rest of the options.
Asking users about features will prompt them to talk to you about features.
If you want to find out what users really want then you are talking about understanding their goals and motivations. I've found the easiest way to start doing this is user interviews, not about features but about how users use your product and products like it, why they are using it and how it fits in with their life.
Once you build an understanding of what your users are trying to do with your product and why they want to do it you are in a position to make an informed judgment as to whether the features people requested are what they really need.
Ideally I think your problem is about understanding users rather then just listening to their requests.
This is an old question with a lot of good answers already, but I thought I'd just add a little bit of personal experience for the sake of people who end up here in the future through a search like I did.
If your project does not need to gain an audience as quickly as possible in order to succeed (like a webapp) if it's more of an internal project or product to be sold for a fixed client, or type of client, then I believe your best bet is to go the 37signals way: give your users the absolute minimum they need in order to accomplish the most basic tasks of the most basic cycle of work at first, then listen to what they say it's objectively missing in order for them to do their work properly. Not what they want or would like it to have, but what they really need. And the only way you know for sure what you really need is when you don't have it.
I worked as the designer in the development team of an intranet-based "heart-of-the-company" app that followed that strategy, and the results were wonderful. First week: everyone was pissed. When it was over, 90%+ of approval, and the app was still simple and beautiful. And most of the people who were not entirely satisfied seemed to understand why it couldn't be like they wanted, and the main request of nearly everyone was to, whatever we did, keep the app simple.
Again, if you're working on a product or website that needs to attract people first, that might not be feasible or delay things a lot. But if you have some control or leeway over the userbase, I'd definitely recommend this approach.
You don't ask for features. You ask for problems. Pain points. Find out what they hate about their current solution. Find out what eats in to their time.
When you know what they don't like, then you build the solution to those problems.
When you solve real problems, then you're creating real products that people will gladly give you money for.
But what's also important is respecting them during your research phase. Surveys are still great for doing research, but if you ask them a dozen questions, they will hate you. You need to respect their time and use a survey tool that engages them and leaves a great impression.
It's a proven fact users don't know what they want. What you need to ask them is what is wrong with what there is now - what problems are they having with your software? why aren't they using x feature and y control? why interaction x worked for them while interaction y made them try to gauge their eyes out?
Of course to be able to ask those questions, you need to do some field study and see what features are used, what patterns your users exhibit and analyze that data. That analysis will give you the base for much more specific questions which users are able to answer decisively and accurately.
If you're serious, you videotape them at their work, and then you break down what they are trying to accomplish and how your product can help them. This is part of a whole discipline called usability engineering. A good introduction to technique is Jakob Nielsen's book Usability Engineering. Back before he became a shameless huckster, Jakob was a very good scientist and he learned a lot about cheap ways of figuring out what users need. Especially good if you're on a budget. What impressed me most was using paper prototypes; this is a great way to mock up software you haven't built yet and helps answer your question about what to build next. Until I saw this technique in action I couldn't believe how effective it could be.
P.S. One example of what happens if you just ask people: 90% of the feature requests for Microsoft Office 2007 were for features that were already in Microsoft Office 2003. In that case what users needed were better ways of finding what was already there. I wish I could find where I read about this... sorry not to have a reference.
I'm assuming based on your wording that you are building a product to sell, and not building something to order for a specific client.
In that context, I'd say that you should start by becoming a user yourself and building the features you need in the way that you want it. As you evolve the product, you'll need feedback from other users, but this at least this gets you started and breaks the chicken-egg cycle.
As for measuring actual usage of features, you can set up a discussion forum to get feedback on the features you added... you don't need anything too complicated if you are time-strapped.
I personally like the hands off approach from customers. They give you high level requirements and you provide the implementation. Your software team/company/division are supposed to be the experts. Sure you will make some mistakes, if its horrible the customer will pipe up and you will fix it, but generally having the implementation up to you and your developers is a fun dilemma to solve.
Research, research, research. Learn from others designs, then make your own kickass design. Not easy but then again they don't pay developers the big bucks for nothing.
That's a good question.
If you're building an FPS game, you really need to know for yourself what should be included, because 99% of your users will never contact you to say "I wish your game just had X". An experienced beta-testing team can help here.
If you're writing an accounting application, you need to understand the industry and what users are trying to accomplish when they use your product, and try and focus your feature set around those goals.
If you're writing a custom app for 100 users in one business, you could have a chat to the dozen or so most avid users of the software. They're the ones who know all the forms back-to-front, have discovered all the undocumented shortcut keys, and have also figured out how to circumvent many of your data validation rules.
Imagine you are them
Use Cases.
What will they do with that feature?
It works like this.
People take actions. We build software to help them take actions
In order to take an action a person must make a decision. We build software to help them make decisions.
In order to make a decision to take an action, a person needs information. We build software to collect and present information.
Every feature must be an Action, a Decision or Information. And the connection had better be direct. Information that does not lead to a decision or an action isn't even "nice to have" -- it's junk.
Users say a lot of things. What do they do? What decisions do they make? What information do they need?
Edit
Note that not everyone is good at describing use cases. Some people have no vision and will simply tell you what they do today without understanding how they are creating business (or personal) value. They may not really know what decisions they're supposed to be making, and are vague on the information they need.
Other users know what value they create, and why, and can discuss use cases well. They can envision alternative ways to create value; they can articulate options for their actions. Decisions don't have a lot of alternative implementations (people make decisions, not software) and the information required doesn't change much, either.
Watch them.
Identify bottlenecks in their work
Create something that solves that
bottleneck in an elegant way
Let them use it
Repeat until everyone is happy
Based on the principles:
Users know what they want, but they
don't know what they really need.
You ARE
NEVER going to get it right the
first time.
It seems like a chicken-and-egg problem. Much like computing PageRank. A page's page rank is dependent on the PageRank of other pages linking to that page.
One way of computing PageRank is by iteration.
Iteration is the key!
A. Voting
Gather a biiiig list of features all users want (make them enumerate each feature they want).
Then have them review the list and allow them to vote on features. Say, give em 100 points to distribute on features. They can give more than 1 point to a feature.
B. Analysis
Analyze the business model, List the features that you think is needed.
This is needed because:
users sometimes don't get the big
picture
you have this REALLY great
idea that users won't think of in a
bajillion years.
C. Implement
Analyze list from A and B, merge, remove a few, improve some. Implement.
D. Test
Test it on users. Hear their complaints. Look at
- features they use often
- stuff they get stuck on
- etc etc etc
E. Iterate
Usually, users do not always know what they want and whether they want anything. In our company sales people go to existing and potential customers, show them our product and explain them why they desperately want that.
In my time in university we were taught something called "userp-driven development". Here you really have to go to the customer, observer how people there work, what tools do they use, and try to find out what could facilitate their life. You then create a mock-up, go to the customer again, present it to the users, get their feedback and then proceed to improve your mock-up. When everyone more or less agrees to the course of action, you do implementation, regularly showing the customer what you have trying to get correction feedback as early as possible.
Important is not to talk to the managers who want the product, but to the users who will use the product. Otherwise the whole play will bring you nothing.
P.S. Asking them directly "What do you want?" could be a dangerous question...
Babylon 5 - What do you want?
It's called Market Research.
No, this wasn't a dig at the guy, that's really what it is about. Sure, there's a bunch of techniques that UCD people use in the field to get user requirements, but they are exactly the same tools used by market researchers. Card Sorting, Priority lists and so on are all market research terms.

Resources