1: \begin{abstract}

2: Exploring statistics of locally

3: connected subgraph patterns (also known as network motifs)

4: has helped researchers better understand the structure and function of

5: biological and online social networks (OSNs).

6: Nowadays the massive size of some critical networks -- often

7: stored in already overloaded relational databases -- effectively limits

8: the rate at which nodes and edges can be explored, making it

9: a challenge to accurately discover subgraph statistics.

10: In this work, we propose {\em sampling methods} to accurately

11: estimate subgraph statistics from as few queried nodes as possible.

12: We present sampling algorithms that efficiently and accurately

13: estimate subgraph properties of massive networks.

14: Our algorithms require no pre-computation or

15: complete network topology information. At the same time,

16: we provide theoretical guarantees of convergence.

17: We perform experiments using widely known data sets,

18: % and real networks

19: %Experimental results show that,

20: and show that for the same accuracy, our algorithms require an order of magnitude

21: less queries (samples) than the current state-of-the-art algorithms.

22: \end{abstract}

23: