"... In PAGE 9: ... The generated databases thus fulfil the support differ- ence demands we formulated in Equation 3. The third column of Table2 contains the average supports of item sets that are newly found in the gener- ated databases; these supports are very low. All this together clearly shows that there is a large pattern- similarity, thus showing a high quality according to our problem statement.... ..."

"... In PAGE 3: ... It enumerated all 4 families of patterns: those that begin with A, C, G and T. We show in Table1 the results for a family of patterns that start with A. Page 3 of 18 (page number not for citation purposes) ilarly ranked scores from simulations.... In PAGE 4: ... New pat- terns are checked against existing ones to avoid redun- dancy. In our example (see Table1 ), the pattern A.A can be combined with AA, AG and AT but the list of positions is null for AAA and contains only one position for AGA.... ..."

"... In PAGE 3: ...1 FrequentPattern Tree To design a compact data structure for e#0Ecient frequent pattern mining, let apos;s #0Crst examine a tiny example. Example 1 Let the transaction database, DB, be #28the #0Crst two columns of#29 Table1 and the minimum support... ..."

"... In PAGE 7: ... We use the S amp;P 500 index data set for our clustering experiments. Table5 shows the dramatic increase in the number of frequent patterns as the minimum support threshold is de- creased. As can be seen, the number of frequent patterns increases up to 11,486,914 when we reduce the support threshold to 1%.... ..."

"... In PAGE 7: ... We use the S amp;P 500 index data set for our clustering experiments. Table5 shows the dramatic increase in the number of frequent patterns as the minimum support threshold is de- creased. As can be seen, the number of frequent patterns increases up to 11,486,914 when we reduce the support threshold to 1%.... ..."

"... In PAGE 9: ... 5.3 Efficiency Results Table4 shows the number of frequent patterns (rules) mined by XMiner, and time for training and testing. The results underscore the high e ciency of that XMiner (XM) engine.... ..."

"... In PAGE 3: ...rs are rarely encountered. For example, NPs can exhibit di erent patterns, e.g. NP!PRO, or NP!DET N, both of which are very frequent patterns, but also, less frequently, NP! DET ADJP N N. Table1 shows the ve most frequent patterns for each type, together with their frequency of occurrence in the training corpus. In order to maximize the time e ectiveness of the bilingual speaker who will translate the corpus, we wish to focus on those patterns that occur frequently.... ..."

"... In PAGE 4: ...ivided into 32-bit words (e.g., 16 words for a 64-byte line). Each 32-bit word is encoded as a 3-bit prefix plus data. Table1 shows the different patterns corresponding to each prefix. Each word in the cache line is encoded into a compressed format if it matches any of the patterns in the first six rows of Table 1.... In PAGE 4: ...ivided into 32-bit words (e.g., 16 words for a 64-byte line). Each 32-bit word is encoded as a 3-bit prefix plus data. Table 1 shows the different patterns corresponding to each prefix. Each word in the cache line is encoded into a compressed format if it matches any of the patterns in the first six rows of Table1 . These patterns are: a zero run (one or more all-zero words), 4-bit sign-extended (including one-word zero runs), one byte sign-extended, one halfword sign-extended, one halfword padded with a zero halfword, two byte- sign-extended halfwords, and a word consisting of repeated bytes (e.... In PAGE 9: ...3 Which Patterns Are Frequent? Frequent Pattern Compression (FPC) is built on the observation that some word patterns are more frequent than oth- ers. We experimented with cache snapshots for our different benchmarks to come up with a reasonable set of frequent patterns (described in Table1 ). Figure 4 shows the relative frequency of incompressible words, zero words and words compressible to 4, 8 and 16 bits.... ..."