I have stored workflows. They're trees with decision points. Basically every point in the data is a command issued. Adding it all up is a workflow to build something from the commands.
I'm trying to use azure ml to take a partially completed workflow from a user and match it against these stored workflows.
Adding to the difficulty is that I'm never sure when the user has started or stopped a workflow so it's always a Time preference match that is never perfect.
Despite days of searching I can't find any information on canned algorithms that do this type of pattern matching.
Can someone suggest where I might find some information on taking a data series (not numeric) and match that to a tree graph of similar values in real time?
Related
Can chatbots like [Rasa] learn from the trusted user - new additional employees, product ids, product categories or properties - or unlearn when these entities are no longer current ?
Or do I have to go through formal data collection, training sessions, testing (confidence rates > given ratio), before the new version be made operational.
If you have entity values that are being checked against a shifting list of valid values, it's more scalable to check those values against a database that is always up to date (e.g. your backend systems probably have a queryable list of current employees). Then if a user provides a value that used to be valid and now isn't, it will act the same as if a user provided an invalid value in the first place.
This way, the entity extraction can stay the same regardless of if some training examples go out of relevance -- though of course it's always good to try to keep your data up to date!
Many Chatbots do not have such a function. Except avanced ones like Alexa, with the keyword "Remember" available 2017 +/-. The user wants Alexa to commit to memory certain facts.
IMHO such a feature is a mark of "intelligence". It is not trivial to implement in ML systems where coefficients in their neural network models are updated by back-propagation after passing learning examples. Rule-based systems (such as CHAT80 a QA system on geography) store their knowledge in relations that can be updated more transparently.
I have a data set of about 1 million employer names. These names are from a free-form text field so they include misspellings and variations in the way they are inputted (e.g. "Amazon" .. "Amzaon" .. "Amazon.com" .. "Amazon Web Services" .. "AWS").
I want to either A) group these 1 million so I have a somewhat accurate sense of how many unique employers are in the data set or B) be able to find all variations of any given employer.
So far, I've been using the data in Tableau, then filtering on "employer name" and searching all variations of the name I can think of. But it's tedious and I'm pretty sure I'm leaving many out.
I've also used the fuzzy add-in for excel but it hasn't worked that well on misspellings, special characters...
Tableau just isn't suited for doing this kind of analysis straight out of the box, and I would highly recommend doing some pre-processing on your data before putting trying to build a workbook around it.
Like another commenter said, you could look into using Tableau Prep Builder for a one-time transformation on your data set, but if you wanted to automate this process it costs extra to add functionality to whatever Tableau Server installation you have.
If you're familiar with Python or R (and the integration between Tableau Server and those services is supported by your organization), you could look into building a script to run the transformation real-time, but it probably won't be too efficient.
Try experimenting with Tableau Prep Builder - the companion tool that comes with your Tableau Creator license. It has a group feature that is designed for just these problems.
In Prep Builder, you’ll just need to connect to your data, add a cleaning step, and then add a group to your cleaning step.
My friend has a small business where customers order services using email. He receives several emails a day and sorting thru it is becoming cumbersome.
There are about 10 different kind of tasks the customer can request, and for each there are one or two words that specify it. The other info present in the emails is the place where the service is to be delivered, the time, and the involved people's names. The email also contains an ID, a long number with a fairly standard format.
The emails are very unstructured, but all contain the key info above. My question is: what is the best method to sweep thru these emails and extract the key info (such as type of service, place, people's names, the ID etc)?
I thought about some kind of pre-processing, then pass it thru AlchemyAPI and then test the Alchemy output using Neural Networks for each feature (key info). This can be supervised learning as I can do a feedback loop all the time, as once the info is inputted, I can have someone to validate.
Any ideas? Thanks
I guess some parts (ID, task, time) can be captured by a regular expression and dictionary matching. Have a look at GATE's JAPE tool.
It should be fairly easy to assemble a dictionary and then use the lookups for the "task", also you can reuse the available jape rules for date/time and write a new one for the ID (also, a simple regex could be fine).
For matching the location and people's names you should be careful, openCalais and alchemyAPI can give you good results if names and places are used in well defined sentences and will probably make more mistakes with some tabular or weird format. Also you can never be sure you captured the place and person correctly so don't rely on that for processing orders directly.
If you have more information about mails' structure or expected names and places (i.e. you have a "clients" table with all possible names), you would probably want to do your own tagging, otherwise I'd stick to openCalais or alchemyAPI + some regular expressions.
P.S. I assume all mails are in English.
I have a requirement in which I need to import data from excel (CSV) to Dynamics CRM regularly.
Instead of using some simple Data Duplication Rules, I need to implement a point system to determine whether a data is considered duplicate or not.
Let me give an example. For example these are the particular rules for Import:
First Name, exact match, 10 pts
Last Name, exact match, 15 pts
Email, exact match, 20 pts
Mobile Phone, exact match, 5 pts
And then the Threshold value => 19 pts
Now, if a record have First Name and Last Name matched with an old record in the entity, the points will be 25 pts, which is higher than the threshold (19 pts), therefore the data is considered as Duplicate
If, for example, the particular record only have same First Name and Mobile Phone, the points will be 15 pts, which is lower than the threshold and thus considered as Non-Duplicate
What is the best approach to achieve this requirement? Is it possible to utilize the default functionality of Import Data in the MS CRM? Is there any 3rd party Add-on that answer my requirement above?
Thank you for all the help.
Updated
Hi Konrad, thank you for your suggestions, let me elaborate here:
Excel. You could filter out the data using Excel and then, once you've obtained a unique list, import it.
Nice one but I don't think it is really workable in my case, the data will be coming regularly from client in moderate numbers (hundreds to thousands). Typically client won't check about the duplication on the data.
Workflow. Run a process removing any instance calculated as a duplicate.
Workflow is a good idea, however since it is being processed asynchronously, my concern is the user in some cases may already do some update/changes to the data inserted, before the workflow finish working.. therefore creating some data inconsistency or at the very least confusing user experience
Plugin. On every creation of a new record, you'd check if it's to be regarded as duplicate-ish and cancel it's creation (or mark for removal).
I like this approach. So I just import like usual (for example, to contact entity), but I already have a plugin in place that getting triggered every time a record is created, the plugin will check whether the record is duplicat-ish or not and took necessary action.
I haven't been fiddling a lot with duplicate detection but looking at your criteria you might be able to make rules that match those, pretty much three rules to cover your cases, full name match, last name and mobile phone match and email match.
If you want to do the points system I haven't seen any out of the box components that solve this, however CRM Extensions have a product called Import Manager that might have that kind of duplicate detection. They claim to have customized duplicate checking. Might be worth asking them about this.
Otherwise it's custom coding that will solve this problem.
I can think of the following approaches to the task (depending on the number of records, repetitiveness of the import, automatization requirement etc.) they may be all good somehow. Would you care to elaborate on the current conditions?
Excel. You could filter out the data using Excel and then, once you've obtained a unique list, import it.
Plugin. On every creation of a new record, you'd check if it's to be regarded as duplicate-ish and cancel it's creation (or mark for removal).
Workflow. Run a process removing any instance calculated as a duplicate.
You also need to consider the implication of such elimination of data. There's a mathematical issue. Suppose that the uniqueness' radius (i.e. the threshold in this 1D case) is 3. Consider the following set of numbers (it's listed twice, just in different order).
1 3 5 7 -> 1 _ 5 _
3 1 5 7 -> _ 3 _ 7
Are you sure that's the intended result? Under some circumstances, you can even end up with sets of records of different sizes (only depending on the order). I'm a bit curious on why and how the setup came up.
Personally, I'd go with plugin, if the above is OK by you. If you need to make sure that some of the unique-ish elements never get omitted, you'd probably best of applying a test algorithm to a backup of the data. However, that may defeat it's purpose.
In fact, it sounds so interesting that I might create the solution for you (just to show it can be done) and blog about it. What's the dead-line?
I have an app that extracts information from incoming messages. The messages all contain the same information, but they have different forms depending on the source that sent them.
Example:
Message from source A :
A: You spent $50.00 at Macy's on 2/20/12
Message from source B :
Purchase, $50.00, Macy's, 2Feb2012, Balance $5000.00
Every message from a single source has the same form though. So at the moment, I'm doing it by writing a set of regular expressions to first identify which message I'm trying to decode (i.e. what source it came from so I know what the form of the message is), and then extracting the necessary information from the message (in the above example, I want to know the transaction amount, the store where the transaction happened, and the date). If I discover a new source for a message, or a source changes the format of their message (doesn't happen very often, but could happen), I need to manually write the regular expressions for that message. I'm sure however that I could automate this using some kind of machine learning technique. I just don't know much about machine learning, and I don't know where to even start looking for a technique that would apply to my problem. I would like someone to just point me in the right direction on where to start reading.
In order to detect and label amounts, dates, person names and similar information you can use a technique called Named Entity Recognition. The Stanford Named Entity Recognizer comes with pretrained, ready to use models.
You also use whatever labeled data you have generated so far to learn a custom model for your application. The standard techniques used for this purpose are Conditional Random Fields or Sequence Perceptron. There are many toolkits implementing these models, including:
Wapiti - A simple and fast discriminative sequence labelling toolkit.
Sequor - sequence labeler based on Collins's (2002) perceptron.