The CareerAdvisorBot is an AIML chatbot designed to help computer science students identify and recommend a job type or career specialization based on a short survey. The bot will interact with the student asking him/her questions. Once all questions have been answered the bot will evaluate the student's strengths and preferences and provide a personalized job type recommendation.
The following five initial career paths are supported:
- Platform Engineer
- Frontend Engineer
- Backend Engineer
- Network Engineer
- Machine Learning Engineer
The files for this project are stored on pandorabots and published in this repo.
The training cases are the multiple-choice options the user can select from in each survey question. Each option maps to a job type. I chose to store these as AIML maps that way each training case was could be accessed uniformly, easily, and repetitively. It would also make it easier for adding new training cases: Add the new [Key, Value] to each map and add new associated <button> and <category> for each survey question, no need to edit any existing code. There are three maps:
In addition to training cases that help determine the resulting job recommendation, the result.map is used to capture what the message will be to the user once a decision has been made.
The main AIML logic can be found in careeradvisor.aiml.
The training cases were arbitrarily selected to determine a recommendation and are NOT indicative of the best and or right recommendations. The logic could be easily extended for the bot to <learn> the categories based on user responses.
The job type mapped to the user's selected option is saved as a predicate a1 and a2 allowing the chatbot to examine whether these values are matched and remember the user's previously denoted preference. Additionally, if a1 and a2 do NOT match, the tiebreaker1 and tiebreaker2 predicates store the concatenated options from question one and question two based on the user's responses. The two tiebreaker options are presented to the user assuming the job type indicated by questions one and two do NOT match.
- Navigate to Pandorabots Directory.
- Search for
Career Advisor Bot 846292. - Select
Career Advisor Bot 846292, which should be the first option in the list (top left). - Click on the orange Pandorabot chat widget icon (lower right) to access the bot.
- Enter any word to start interacting.
The Pandorabot environment makes it easy to get started with AIML and developing chatbots. There is lots of freely available documentation and resources to help aid a first-time AIML developer like myself. The simple interface with common, ready-to-use code snippets made coding in XML a little less painful than with a text editor. And being able to save, run, test, and refresh the predicates in the same window made the testing loop super quick and useful for iterative development.
Unfortunately, as with any sandboxed development environment, there are some opinions baked in and missing features that many seasoned developers will miss. For example, the editor's ability to catch XML syntax errors wasn't always very thorough. An extra >> on the end of an XML tag will result in a bug in your chatbot rather than found by your linter. No IntelliSense to give you contextual help means you need to be better at searching out the solution you need. Any programming language extensions that can help AIML with things like looping or nested data structures aren't available. Unless you pay for an account you can't link directly to your bot once it is published to the Pandorabot directory. And last but not least, while there is integration with GH, if I were to push a new copy of my code to this repo, it will remove any files that do NOT exist in Pandorabot. Meaning this README will get erased and have to be recovered from a previous commit.
Now that the chatbot is published on Pandorabot directory the logs for anyone who uses the bot will be available. The logs and feedback gathered from them can be used to improve the questions, training cases, and results. Once the changes to the bot have been made a new version can be published to the directory.
The Panopto video has been uploaded to the Intro to Artificial Intelligence NIP1 | C951 folder.