4.4 Word Rank

View Post

Xiaojun Wang’s paper included word rank on all words in a document, all nouns and verbs in a document, all noun and adjectives in a document, 2 combined word rank and DF: WordRank3DF2 and WordRank4DF1. In this project, tests target on all words’ word rank, nouns and verbs word rank, but nouns and verbs word rank is not included. Word rank based on direct weighted graph is not included. There are WordRank3DF2, WordRank4DF1 and WordRank5DF5, meanwhile, TF is taken into consideration, 3 groups are added: WordRank3TFIDF2, WordRank4TFIDF1 and WordRank5TFIDF5.

WordRank	Google		Yahoo
3	47.00	20.89%	46.00	20.44%
4	73.00	32.44%	68.00	30.22%
5	93.00	41.33%	88.00	39.11%
6	99.00	44.00%	98.00	43.56%
7	119.00	52.89%	119.00	52.89%
8	133.00	59.11%	119.00	52.89%
9	145.00	64.44%	127.00	56.44%
10	152.00	67.56%	133.00	59.11%
11	149.00	66.22%	129.00	57.33%
12	155.00	68.89%	129.00	57.33%
13	155.00	68.89%	129.00	57.33%
14	156.00	69.33%	135.00	60.00%
15	156.00	69.33%	130.00	57.78%
Average	125.54	55.79%	111.54	49.57%

Table4.16

(a) (b)

Figure4.17 success retrieved pages’ counts per 225 pages and corresponding percentage value by undirected graph word rank.

NounsVerbs	Google		Yahoo
3	29.00	12.89%	24.00	10.67%
4	52.00	23.11%	51.00	22.67%
5	72.00	32.00%	67.00	29.78%
6	85.00	37.78%	82.00	36.44%
7	100.00	44.44%	108.00	48.00%
8	110.00	48.89%	109.00	48.44%
9	125.00	55.56%	117.00	52.00%
10	129.00	57.33%	120.00	53.33%
11	134.00	59.56%	121.00	53.78%
12	134.00	59.56%	121.00	53.78%
13	138.00	61.33%	130.00	57.78%
14	136.00	60.44%	123.00	54.67%
15	140.00	62.22%	130.00	57.78%
Average	106.46	47.32%	100.23	44.55%

Table4.17

(a) (b)

Figure4.18 success retrieved pages’ counts per 225 pages and corresponding percentage value by undirected graph nouns and verbs rank.

WR3DF2	Google		Yahoo
3	128.00	56.89%	122.00	54.22%
4	151.00	67.11%	139.00	61.78%
5	165.00	73.33%	148.00	65.78%
6	168.00	74.67%	145.00	64.44%
7	170.00	75.56%	141.00	62.67%
8	169.00	75.11%	141.00	62.67%
9	167.00	74.22%	136.00	60.44%
10	165.00	73.33%	127.00	56.44%
11	165.00	73.33%	125.00	55.56%
12	165.00	73.33%	131.00	58.22%
13	163.00	72.44%	131.00	58.22%
14	160.00	71.11%	132.00	58.67%
15	161.00	71.56%	134.00	59.56%
Average	161.31	71.69%	134.77	59.90%

Table4.18

(a) (b)

Figure4.19 success retrieved pages’ counts per 225 pages and corresponding percentage value by undirected graph WordRank3DF2.

WR4DF1	Google		Yahoo
3	47.00	20.89%	47.00	20.89%
4	73.00	32.44%	68.00	30.22%
5	150.00	66.67%	139.00	61.78%
6	155.00	68.89%	130.00	57.78%
7	159.00	70.67%	135.00	60.00%
8	161.00	71.56%	129.00	57.33%
9	165.00	73.33%	132.00	58.67%
10	167.00	74.22%	138.00	61.33%
11	169.00	75.11%	144.00	64.00%
12	170.00	75.56%	148.00	65.78%
13	170.00	75.56%	154.00	68.44%
14	171.00	76.00%	149.00	66.22%
15	172.00	76.44%	148.00	65.78%
Average	148.38	65.95%	127.77	56.79%

Table4.19

(a) (b)

Figure4.20 success retrieved pages’ counts per 225 pages and corresponding percentage value by undirected graph WordRank4DF1.

WR5DF5	Google		Yahoo
3	128.00	56.89%	121.00	53.78%
4	151.00	67.11%	141.00	62.67%
5	157.00	69.78%	140.00	62.22%
6	168.00	74.67%	144.00	64.00%
7	167.00	74.22%	142.00	63.11%
8	169.00	75.11%	140.00	62.22%
9	168.00	74.67%	142.00	63.11%
10	167.00	74.22%	134.00	59.56%
11	165.00	73.33%	142.00	63.11%
12	163.00	72.44%	131.00	58.22%
13	163.00	72.44%	137.00	60.89%
14	163.00	72.44%	131.00	58.22%
15	163.00	72.44%	131.00	58.22%
Average	160.92308	71.52%	136.6154	60.72%

Table4.20

(a) (b)

Figure4.21 success retrieved pages’ counts per 225 pages and corresponding percentage value by undirected graph WordRank5DF5.

WR3TFIDF2	Google		Yahoo
3	38.00	16.89%	38.00	16.89%
4	55.00	24.44%	47.00	20.89%
5	85.00	37.78%	86.00	38.22%
6	91.00	40.44%	85.00	37.78%
7	120.00	53.33%	92.00	40.89%
8	124.00	55.11%	99.00	44.00%
9	139.00	61.78%	111.00	49.33%
10	146.00	64.89%	114.00	50.67%
11	145.00	64.44%	119.00	52.89%
12	155.00	68.89%	129.00	57.33%
13	152.00	67.56%	123.00	54.67%
14	159.00	70.67%	136.00	60.44%
15	160.00	71.11%	131.00	58.22%
Average	120.69231	53.64%	100.7692	44.79%

Table4.21

(a) (b)

Figure4.22 success retrieved pages’ counts per 225 pages and corresponding percentage value by undirected graph WordRank3TFIDF2.

WR4TFIDF1	Google		Yahoo
3	48.00	21.33%	48.00	21.33%
4	69.00	30.67%	64.00	28.44%
5	82.00	36.44%	72.00	32.00%
6	96.00	42.67%	84.00	37.33%
7	110.00	48.89%	102.00	45.33%
8	126.00	56.00%	111.00	49.33%
9	137.00	60.89%	125.00	55.56%
10	143.00	63.56%	128.00	56.89%
11	156.00	69.33%	136.00	60.44%
12	158.00	70.22%	139.00	61.78%
13	160.00	71.11%	131.00	58.22%
14	161.00	71.56%	129.00	57.33%
15	159.00	70.67%	132.00	58.67%
Average	123.46	54.87%	107.77	47.90%

Table4.22

(a) (b)

Figure4.23 success retrieved pages’ counts per 225 pages and corresponding percentage value by undirected graph WordRank4TFIDF1.

WR5TFIDF5	Google		Yahoo
3	38.00	16.89%	38.00	16.89%
4	55.00	24.44%	47.00	20.89%
5	78.00	34.67%	67.00	29.78%
6	92.00	40.89%	85.00	37.78%
7	105.00	46.67%	97.00	43.11%
8	124.00	55.11%	99.00	44.00%
9	128.00	56.89%	106.00	47.11%
10	139.00	61.78%	107.00	47.56%
11	144.00	64.00%	114.00	50.67%
12	143.00	63.56%	122.00	54.22%
13	146.00	64.89%	124.00	55.11%
14	155.00	68.89%	129.00	57.33%
15	156.00	69.33%	123.00	54.67%
Average	115.62	51.38%	96.77	43.01%

Table4.23

(a) (b)

Figure4.24 success retrieved pages’ counts per 225 pages and corresponding percentage value by undirected graph WordRank5TFIDF5.

The above charts show the following 2 facts:

1. The start up success retrieval rate increases when the DF takes more parts in WordRankxDFy or WordRankxTFIDFy. This acts very similarly as the basic methods in section 4.1, like TFxDFy or TFIDFxDFy.

2. After exceeding 10 words in a query, the success retrieval rates tend to be flat and stable. This also acts very similarly like the basic methods.

Figure4.25 all word rank related methods comparison

The average success retrieve percentage rates are shown in Figure4.25. WR3DF2 and WR5DF5 show the 2 best results than other word rank related methods. Google’s results are more than 70% and Yahoo’s results are more than 60%.

posted on 2009-06-18 11:44 JosephQuinn 阅读(397) 评论(0) 编辑收藏所属分类: My Master-degree Project

新用户注册刷新评论列表


只有注册用户登录后才能发表评论。




网站导航: 博客园 IT新闻 Chat2DB C++博客博问管理
相关文章: Appendix B 5 Conclusion 4.7 Sentence Rank on Yahoo News Page 4.6 Sentence Rankv 4.5 Random pick sentence 4.4 Word Rank 4.3 Google search tips: meta keys and meta description 4.2 Title 4.1 The basics 3.5 Deep Web Search Engine

Avenue U

常用链接

留言簿

随笔分类

随笔档案

Core Java

最新随笔

搜索

最新评论

阅读排行榜

评论排行榜

View Post

4.4 Word Rank