| D09-1026 | appropriately normalizing the | topic assignments | z . It should now be apparent |
| D10-1005 | probability distribution Bd over | topic assignments | . Each Bd is a vector of length |
| A92-1025 | complete list of industry and | topic assignments | currently in use to categorize |
| D09-1146 | each collection by considering | topic assignments | of documents within each collection |
| D08-1035 | , and write zt to indicate the | topic assignment | for sentence t . The observation |
| A88-1002 | consistent in the way they made their | topic assignments | . The news stories themselves |
| D09-1163 | words and . denotes the vector of | topic assignment | except the considered word at |
| D10-1005 | least squares regression using the | topic assignments | zd to predict yd . Prediction |
| D10-1025 | constraints on the expectations of | topic assignments | to two corresponding documents |
| D09-1092 | for each language l , a latent | topic assignment | is drawn for each token in that |
| A92-1025 | most obvious differences between | topic assignment | and industry assignment are : |
| D10-1025 | term , which tries to make the | topic assignments | of corresponding documents close |
| D10-1005 | response variable depends on the | topic assignments | of a document , the conditional |
| A92-1025 | industry categories . Performance on | topic assignment | was generally much higher using |
| A92-1025 | system used linguistic methods for | topic assignment | and statistical methods for industries |
| D09-1092 | expect that simple analysis of | topic assignments | for sequential words would yield |
| D09-1092 | training tuples and inferring latent | topic assignments | for test documents . These tasks |
| D10-1025 | document is close to the best | topic assignment | of the foreign document . This |
| D10-1025 | instead of making sure that the best | topic assignment | for the English document is close |
| A92-1025 | with coverage of over 90 % on | topic assignment | and performance better than human |
