List of Listings

Chapter 2. Introducing recommenders

Listing 2.1. Recommender input file, intro.csv

Listing 2.2. A simple user-based recommender program with Mahout

Listing 2.3. Configuring and running an evaluation of a recommender

Listing 2.4. Configuring and running a precision and recall evaluation

Listing 2.5. User 5’s preferences in the test data set

Listing 2.6. Changing the evaluation program to run a SlopeOneRecommender

Chapter 3. Representing recommender data

Listing 3.1. Setting preference values in a PreferenceArray

Listing 3.2. Defining input data programmatically with GenericDataModel

Listing 3.3. Triggering a refresh of a recommender system

Listing 3.4. Configuring a JNDI DataSource in Tomcat

Listing 3.5. Configuring a DataSource programmatically

Listing 3.6. Creating and evaluating with Boolean data

Listing 3.7. Evaluating precision and recall with Boolean data

Chapter 4. Making recommendations

Listing 4.1. Revisiting a simple user-based recommender system

Listing 4.2. Updating listing 4.1 to use a custom DataModel for GroupLens

Listing 4.3. Running an evaluation on the simple recommender

Listing 4.4. A simple recommender input file

Listing 4.5. Employing caching with a UserSimilarity implementation

Listing 4.6. The core of a basic item-based recommender

Listing 4.7. Selecting no weighting with a SlopeOneRecommender

Listing 4.8. Creating a JDBC-backed DiffStorage

Listing 4.9. Deploying KnnItemBasedRecommender

Listing 4.10. Creating a cluster-based recommender

Chapter 5. Taking recommenders to production

Listing 5.1. Limiting memory consumed by MemoryDiffStorage

Listing 5.2. A gender-based item similarity metric

Listing 5.3. Example IDRescorer that omits out-of-stock books and boosts a genre

Listing 5.4. Gender-based rescoring implementation

Listing 5.5. Complete recommender implementation for Líbímseti

Listing 5.6. Anonymous user recommendation for Líbímseti

Listing 5.7. Output of a GET to RecommenderServlet

Chapter 6. Distributing recommendation computations

Listing 6.1. A mapper that parses Wikipedia link files

Listing 6.2. Reducer which produces Vectors from a user’s item preferences

Listing 6.3. Mapper component of co-occurrence computation

Listing 6.4. Reducer component of co-occurrence computation

Listing 6.5. Wrapping co-occurrence columns

Listing 6.6. Splitting user vectors

Listing 6.7. Computing partial recommendation vectors

Listing 6.8. Combiner for partial products

Listing 6.9. Producing recommendations from vectors

Chapter 7. Introduction to clustering

Listing 7.1. Sample input for our first clustering example

Listing 7.2. The Hello World clustering code

Chapter 8. Representing data

Listing 8.1. Creating vectors of various apples

Chapter 9. Clustering algorithms in Mahout

Listing 9.1. In-memory clustering example using the k-means algorithm

Listing 9.2. The k-means clustering job entry point

Listing 9.3. In-memory example of the canopy generation algorithm

Listing 9.4. News clustering using canopy generation and k-means clustering

Listing 9.5. A custom Lucene analyzer that filters non-alphabetic tokens

Listing 9.6. In-memory example of fuzzy k-means clustering

Listing 9.7. News clustering using fuzzy k-means clustering

Listing 9.8. Dirichlet clustering using normal distribution

Chapter 10. Evaluating and improving clustering quality

Listing 10.1. Top 10 terms using k-means clustering with a Euclidean distance measure

Listing 10.2. Top 10 terms using k-means clustering and a cosine distance measure

Listing 10.3. Calculating inter-cluster distances

Listing 10.4. A custom Lucene Analyzer that wraps around StandardTokenizer

Listing 10.5. MyAnalyzer with lowercase filter

Listing 10.6. A custom Lucene Analyzer using multiple filters

Listing 10.7. Modifying NewsKMeansClustering.java to use MyAnalyzer

Listing 10.8. A modified cosine distance measure

Chapter 12. Real-world applications of clustering

Listing 12.1. Group-by-field mapper

Listing 12.2. Group-by-field reducer

Listing 12.3. Lucence Analyzer class optimized for Tweets

Listing 12.4. Output artists from the data set

Listing 12.5. Grouping by artist and outputting a unique key-value pair for an artist

Listing 12.6. Using the unique artist names to create a unique integer ID for each artist

Listing 12.7. Using integer ID mapping of the artist to convert tags into a vector

Listing 12.8. Grouping partial vectors of tags and summing them up into full vectors

Chapter 14. Training a classifier

Listing 14.1. Tokenizing and vectorizing text

Listing 14.2. Parsing the data

Chapter 15. Evaluating and tuning a classifier

Listing 15.1. Passing data for AUC metric classes

Listing 15.2. Building a confusion matrix

Listing 15.3. A complete program to train a model with progress diagnostics

Listing 15.4. Model dissection for 20 newsgroups example

Chapter 16. Deploying a classifier

Listing 16.1. Code to parse and encode CSV data

Listing 16.2. Code for byte-level CSV parsing

Listing 16.3. Direct value encoding

Listing 16.4. Main program for the classification server

Listing 16.5. The watcher object loads the model and sets model status

Listing 16.6. Implementation of classification service

Listing 16.7. Accessing the sample classification server

Chapter 17. Case study: Shop It To Me

Listing 17.1. Sample encoder showing one implementation of an interaction encoder

Listing 17.2. Using caching and partial model evaluation to speed up Item selection

Appendix B. Mahout math

Listing B.1. Computing Vector operations efficiently

Listing B.2. Computing aggregations efficiently

Listing B.3. Multiplying a matrix and a vector