The Air Force contracts a variety of positions, from Information Technology to maintenance services. There is currently no automated way to verify that quotes for services are reasonably priced. Small training data sets and word sense ambiguity are challenges that such a tool would encounter, and additional semantic information could help. This thesis hypothesizes that leveraging a semantic network could improve text-based classification. This thesis uses information from ConceptNet to augment a Naive Bayes Classifier. The leveraged semantic information would add relevant words from the category domain to the model that did not appear in the training data. The experiment compares variations of a Naive Bayes Classifier leveraging semantic information, including an Ensemble Model, against classifiers that do not. Results show a significant performance increase in a smaller data set but not a larger one. Out of all models tested, an Ensemble Based Classifier performs the best on both data sets. The results show that ConceptNet does not add enough new or relevant information to affect classifier performance on large data sets.