List of Tables

1. Typical system parameters in 2007. The seek time is the time needed to position the disk head in a new position. The transfer time per byte is the rate of transfer from disk to memory when the head is in the right position.
2. Collection statistics for Reuters-RCV1. Values are rounded for the computations in this book. The unrounded values are: 806,791 documents, 222 tokens per document, 391,523 (distinct) terms, 6.04 bytes per token with spaces and punctuation, 4.5 bytes per token without spaces and punctuation, 7.5 bytes per term, and 96,969,056 tokens. The numbers in this table correspond to the third line (``case folding'') in icompresstb5.
3. The five steps in constructing an index for Reuters-RCV1 in blocked sort-based indexing. Line numbers refer to Figure 4.2 .
4. Collection statistics for a large collection.
5. The effect of preprocessing on the number of terms, nonpositional postings, and tokens for Reuters-RCV1. ``$\Delta\%$'' indicates the reduction in size from the previous line, except that ``30 stop words'' and ``150 stop words'' both use ``case folding'' as their reference line. ``T%'' is the cumulative (``total'') reduction from unfiltered. We performed stemming with the Porter stemmer (Chapter 2 , page 2.2.4 ).
6. Dictionary compression for Reuters-RCV1.
7. Encoding gaps instead of document IDs. For example, we store gaps 107, 5, 43, ..., instead of docIDs 283154, 283159, 283202, ... for computer. The first docID is left unchanged (only shown for arachnocentric).
8. Some examples of unary and $\gamma$ codes. Unary codes are only shown for the smaller numbers. Commas in $\gamma$ codes are for readability only and are not part of the actual codes.
9. Index and dictionary compression for Reuters-RCV1. The compression ratio depends on the proportion of actual text in the collection. Reuters-RCV1 contains a large amount of XML markup. Using the two best compression schemes, $\gamma$ encoding and blocking with front coding, the ratio compressed index to collection size is therefore especially small for Reuters-RCV1: $(101+7.9)/3600 \approx 0.03$. $(101+5.9)/3600 \approx 0.03$.
10. Two gap sequences to be merged in blocked sort-based indexing
11. Cosine computation for Exercise 6.4.4 .
12. Calculating the kappa statistic.
13. INEX 2002 collection statistics.
14. INEX 2002 results of the vector space model in Section 10.3 for content-and-structure (CAS) queries and the quantization function Q.
15. A comparison of content-only and full-structure search in INEX 2003/2004.
16. Data for parameter estimation examples.
17. Training and test times for NB.
18. Multinomial versus Bernoulli model.
19. Correct estimation implies accurate prediction, but accurate prediction does not imply correct estimation.
20. A set of documents for which the NB independence assumptions are problematic.
21. Critical values of the $\chi ^2$ distribution with one degree of freedom. For example, if the two events are independent, then $P(X^{\kern .5pt2}>6.63)<0.01$. So for $X^{\kern .5pt2}>6.63$ the assumption of independence can be rejected with 99% confidence.
22. The ten largest classes in the Reuters-21578 collection with number of documents in training and test sets.
23. Macro- and microaveraging. ``Truth'' is the true class and ``call'' the decision of the classifier. In this example, macroaveraged precision is $[10/(10+10)+90/(10+90)]/2 = (0.5+0.9)/2=0.7$. Microaveraged precision is $100/(100+20)\approx 0.83$.
24. Text classification effectiveness numbers on Reuters-21578 for F$_1$ (in percent). Results from Li and Yang (2003) (a), Joachims (1998) (b: kNN) and Dumais et al. (1998) (b: NB, Rocchio, trees, SVM).
25. Data for parameter estimation exercise.
26. Vectors and class centroids for the data in Table 13.1 .
27. Training examples for machine-learned scoring.
28. Some applications of clustering in information retrieval.
29. The four external evaluation measures applied to the clustering in Figure 16.4 .
30. Comparison of HAC algorithms.