Information Filtering (IF) is an area of research where only a few documents arernselected from a large collection in a dynamic flow of information. Particularly, filtering outrnnews items from a collection has paramount importance in order that a news reader canrneasily find what he/she likes to read. Several research projects are underway tornimplement such a system. rnCollaborative filtering aims at learning predictive models of user preferences, interests orrnbehavior from community data, e.g., a database of available user preferences. It isrncomplementary to content-based filtering and retrieval that is mainly built on thernfundamental assumption that users are able to formulate queries that express theirrninterests or information needs in terms of intrinsic features of the items sought.rnMany newspapers are being published in Amharic. Almost none of them use automatedrnsystems for filtering news. With the increasing number of such news, it is evident that arnlot of textual information is accumulated which makes it difficult for the reader to find fewrndesired news from a collection. It was felt that an experiment should be underway tornextract such news on the basis of collaborative interest.rnThe purpose of this research was, therefore, to explore the potential application ofrnArtificial Neural Networks (ANN) for filtering Amharic news based on preferences ofrnreaders. The Back propagation (BPN) and Self Organizing Map (SOM) algorithms werernused to develop a model for Amharic news filtering where news items were selectedrnfrom two popular Amharic newspapers. The preferences of reading these items,rncollected from active readers of the newspapers, were used to develop the first modelrnwhereas the weighted term-by-document matrix of the news items in the sample wasrnused to classify the news items.rnThe experiment was undergone in twofold; developing a model for predicting userrnpreferences of reading news items and classifying the news items in the sample tornpredefined categories. The results showed that ANNs can be used to model userrnpreferences of news items written in Amharic. The Mat lab neural network toolbox wasrnused to develop both models.rnThe result indicated that with Model 1, containing the preference list, it could predictrn83.3% of the preferences in the training set and 79.8% of the preferences in the test set.rnThat is, a news item is likely to satisfy the readers in the test set 79.8% of the time.rnModel 2, the Self-Organizing Map (SOM) model, was also trained so as to classify newsrnitems into each category of the news. The best model could classify the items 76.5% ofrnthe time. 72.9% of the news items in the test set were correctly classified into thernrespective category. However, as neural networks learn from large examples, extendedrnresearch is recommended.