Knowledge-based search

Now & Then combines Bayesian probability with iterative retrieval to determine each query's most likely context and meaning.

It also combines its understanding of your personal syntax. your worldview, your query history, your most likely areas of interest, and (if needed) iterative relevance feedback, to infer what information will most likely satisfy each query.

In other words, the more Now & Then knows about your personal domain (worldview, words used, current interests, etc.), the better it can find and retrieve relevant information (and disregard non-relevant information).

Say, or example, your recent journal entries describe your planned vacation to the Cayman Islands. You entered information on all of the sight seeing activities that showcase island marine life. Because Now & Then "understands" the part of your worldview on which you are currently focused, your queries regarding "Turtle songs" and "average dolphin weight" will retrieve information relevant to your vacation plans―not the information you entered years ago about a British pop music band and last year about having a great time at the Dolphins-Broncos football game.

The search algorithm uses a Bayesian decision tree that favors the information most likely to satisfy your query. If it cannot resolve an ambiguity, Now & Then may display a list from which you select a term that "nudges" the search in the correct direction (e.g. "Does Dolphins refer to football" or "mammals?"). For a complex Query, your selections from the term list may be used with multiple search iterations to guide the search toward relevant, and away from, non-relevant journal information.

Natural Language Query
A convenient way to retrieve information from your journal is to simply enter a natural language question, phrase, or word. If you wish, you can include details in your queries that describe things like categories, dates, date ranges, people and places.

Now & Then does a good job of parsing your query to "understand" concepts such as dates, topics, and the knowledge of all information stored as part of your worldview. Because your queries launch contextual-aware searches, you can be creative and have fun with your journal searches:

• "where did I spend my birthdays as a teenager?"

• "What are the weddings (parties, football games, funerals, company functions) I attended in the 1990's?"

• "Where was I, what was I doing, and on what was I focused on this date every year since 1982?"

• or play WADAYAM ― "What A Difference A YeAr Makes". See how today compares with the same day one year ago (variations 5 years ago, 10 or 30 years ago)

Prompted Search
If you find that you are repeatedly entering the same queries, you can save any query by creating a stored Prompted Search, or a Search Template for future use. With Prompted Searches and Templates you use "drop down" selection boxes, "fill-in" fields or answer prompts. These tools are handy if your searches include multiple search criteria such as specific Categories, Topics, dates & ranges, or recurring concepts and phrases. Both tools can be iteratively used to progressively refine and narrow your search.

Template-based searches are particularly useful in retrieving the more structured information that you track and monitor. To track your exercise program, you might create a Template with fields for "Minutes Walked', "Distance Walked", "Walking Pace" (automatically calculated), "Walking course", "Energy Level", "Observations", "date or range", and "mini graph" (automatically calculated).

Also see The Intelligent Journal:
Context Aware Search   ▪   Query interpretation technologies  

home