Automating the publishing of puzzle books such as Sudoku and Word Search: 

Problem statement: Puzzle books sell for timepass activities. They are considered commodity items, are highly labor-intensive, available to purchase from a store or even download for free, widely varying in quality and content, and usually have a poor retention span. How do we make it easy to generate, tailor, and more interesting puzzles for the end-user? Can we disrupt this staid market with logic and automation? This article attempts to elaborate on an affirmative note.  

Solution: There are a few interesting patterns to observe with puzzles such as Sudoku and Word Search:  

1. They have definite unequivocal answers which are easy to tally against, once the solving is done.  

2. The same solution can be used to create many puzzles simply by selective hiding or nominal changes in positions. A sudoku key can be used to cut different puzzles by hiding different positions. A Word search can realign different orientations of the hidden words without replacing them to create different puzzles. The automation for generating different puzzles depends only on randomizing the selection which is easily programmable. Refer to the sample application program included for Sudoku and Word Search in the references.  

3. The content of the word search is a potpourri of words from a dictionary. This is hardly interesting to the end-user. On the other hand, the domain of a person’s profession or her interests leads to more enticing puzzles to solve. The candidate words for a puzzle can also be mined from the person’s handheld device or data source which is both temporally and spatially closer to the user’s interests. Even proper nouns such as names of friends, things, or places can be included in the word search.  

4. Puzzles available to download from an Application Store on a mobile device compete with one another on the look and the feel of the puzzles. If they used content from a source that is specific to the user, it would be disruptive to the market. A users’ interests can be learned with natural language processing from the public web pages visited. If the user allows access to files on her device such as downloads, then they provide a better data source. Data may arrive continuously which allows us to batch until enough candidates have been collected.  

5. The organization of candidates for both kinds of puzzles – Sudoku and Word Search can be such that they are done once and stored in a single data store available universally. Then it can be used to dynamically cut the puzzles.  

6. The logic to cut the puzzle and present it can be packaged in an application that can be downloaded, installed, and run on the buyers’ device. The content and the recommendations for the candidates can be organized in the backend which could be hosted in a cloud computing environment. 

Conclusion: The process for creating these puzzles is automatable. Intelligence to present an enticing puzzle to the user can also be incorporated. 

References: 

1. Sudoku: https://github.com/ravibeta/csharpexamples/blob/master/Sudoku/Sudoku/Program.cs