SaffronAPIs

SaffronAPIs for Reasoning Methods

So now that you’ve got a MemoryBase, what can you do with it?

Anything, pretty much. But mainly, you can reason with it. And it all revolves around how you choose to apply Saffron’s REST APIs for Reasoning Methods.

With the Saffron Natural Intelligence Platform, you can query and compute anything you want, and we’ve made it simple for you to invent new analysis methods and new ways of applying the Experience in your data toward solving even your most challenging business problems.

In other words, you can – quite literally – “get outside the box,” and gain remarkable new perspectives.

Applied Reasoning Methods

Reasoning Methods provide application and utility views of SaffronMemoryBase for rapid application development. Following are five of the available views, and Saffron along with its customers and partners continues to add to this list.

Reasoning by Analogies

Analogies – or finding out what is “similar” – is one side of the coin of what’s often called “entity analytics,” or “entity resolution.” It’s useful for finding duplicate IDs, name variants, or intentional aliases. Or, you can use this Reasoning Method to look things up and reason by similarity.

For example: Let’s say you learn of one customer who has complained about a defect in one of your products. And you want to know whether there are any other customers (or other people, places or things) who have or are likely to have the same problem, or a related one. With Saffron, you can quickly find all that out. Or, if you’re trying to close a business deal, you might wish to know whether you have Experience in your data to recall other, similar situations, find out whether those deals closed, and if so, how? What was different or special?

Using Saffron’s Analogies view in situations like these, you can see everything that happened in the past, in context and relevant to the current situation, so you can make a better decision today.

Key Benefits of Reasoning by Analogies:

- Eliminates the need to write matching rules

- Directs SaffronMemoryBase to recall the most informative signature of the category values you provide and uses them to look up other values in the category with a similar signature

- Weighs the factors in the signature by a measure of the information available, or entropy, in the data observed

- Knows that entropy is a measure of information in a frequency distribution; and frequency distributions are defined in SaffronMemoryBase

Reasoning by Connections

Connections is the other side of entity analytics. Connections means finding out what is related, knowing who/what is related to whom/what. As is the case with Analogies, with Reasoning by Connections you can navigate the network of memories yourself. Or, if you want, you can use Connections to provide a higher-level query language for searching and retrieving entities.

Key Benefits of Reasoning by Connections:

- Returns entities according to entity rank rather than documents according to document rank

- Includes a search-engine-like query capability including the AND, OR and NOT operators

- Allows you to specify the category type of entity you want returned if you are looking for specific things such as people or organizations

- Returns entities in rank order of several metrics

- Includes a list and relative ordering metric for simple lists of entities ranked by how they are connected to the query

- Includes more advanced ordering factors for pair or triple connections to the query terms, the counts associated with these connections, and the frequency strength of the connections

Reasoning by Networks

Reasoning by Networks means seeing how entities in a list (a set) are connected to each other. Visualizing a network of entities is another way to analyze them. Rather than a rank-ordered list returned by the Connections view, the Reasoning by Network view shows how entities in a list or set are connected to each other. You may want to define your own functions for networks analysis, and with Saffron, you can.

Key Benefits of Reasoning by Networks:

- Returns not only the connections but also reveals the strength of those connections, between a given list of category values

- Allows you to define two sets: a source set and a destination set

- Returns a bi-partite graph when a second set is included, showing only the connections between the sets

- Supports rapid prototyping while allowing you to build your own network functions as you go

Reasoning by Classifications

Reasoning by Classifications helps you make decisions. The built-in graph structure in SaffronMemoryBase supports schema-free semantics, and Reasoning by Classifications also demonstrates how SaffronMemoryBase supports schema-free statistics.

Key Benefits of Reasoning by Classifications:

- Recalls similar cases and their associated classification to provide an answer, a recommendation

- Uses Experience-based or memory-based reasoning with a “nearest neighbor” classifier

- Allows you to classify a new situation according to associated actions and outcomes from your corpus of past experiences

- Moves you into the world of Experience-based Decision Support.

Reasoning by Temporal Analysis

Temporal Reasoning introduces Trends and Episodes as two powerful approaches for incorporating elapsed time, time ranges, and time sequencing more directly into your decision analysis. With these capabilities you can find trends associated with a given query, for example, or detect changes in sentiment or any other influences you specify by query terms. Plus, you can identity difficult-to-discern, non-obvious patterns you didn’t know were in the data, but may have had a hunch about.

When we think of a stock or group stocks we would like to understand how they are being impacted by what is changing in the world around them, or what trends are showing a connection with them. Or consider a patient’s or group of patients’ experience with a medical treatment protocol, and all of the reactions and experiences, positive and negative, that are occurring during the protocol treatment period or after.

Key Benefits of Reasoning by Trends:

- Identifies, by strength, all of the connections to your query terms over the specified time period

- Provides visibility into emerging or changing patterns happening with the item that you are interested in

- Delivers back far more than just a “histogram” of events over time; you now understand what is happening “around” a person, place, thing or event/issue.

Key Benefits of Reasoning by Episodes:

- Provides insight into historical attribute patterns over given time frames

- Finds repeated, non-obvious patterns

- Retrieves a list of temporal patterns (episodes) to your specified query terms, rank-ordered by a combination of signature and support size

Customer-defined Services

Saffron’s core REST APIs are also available to customers to build their own Reasoning Methods, providing ultimate flexibility for inventing and innovating within your business. Didn’t find what you were looking for above? Then let’s build it!

Summary

When you need to know things, or find a solution to a problem, your mind starts searching for what’s similar or related. In your mind’s eye, you see or visualize how things are connected. As you sort through it all, you find patterns or trends within your memories and their connections.

Saffron REST APIs work much the same way, by applying Natural Intelligence to your data. They work like you do.

So that’s what you can do with a MemoryBase.