NLPIR\/ICTCLA2018 ACADEMIC SEMINAR 8th ISSUE COMPLETED<\/span><\/span><\/b>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f0-y\u001eu7e
\nl#@){7e\u0010H<\/span><\/p>\n

\u00a0<\/span><\/span>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f0&?3n\u0014m-C\u000ee\u0002{3t
\nf\u0006P<\/span><\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 Last week, Shen Li gave a presentation about the paper, Design Challenges and Misconceptions in Neural Sequence Labeling and shared some opinion on it.<\/span><\/span><\/span>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f0\u0001p\u000b~
\n` ^\u0015f\u000em\bQ\u0019J\/M3P+I#L<\/span><\/p>\n

\u00a0<\/span><\/span>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f0
\nk.u\u0010r)R:o4{\u0019i<\/span><\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 After the presentation, several questions were asked. The Q&As are listed as following:<\/span><\/span><\/span><\/p>\n

*T\u0016q\u0019D\u0002b\u001fB,s\u0012\\3d0<\/span><\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 Q: What is oov, oobv and ooev?<\/span><\/span><\/span>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f0,s\u0011h7?
\nY*X,g\u0012S\u0002?H%h\u0010z<\/span><\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 A: oov is Out-of-vocabulary, oobv is out-of-both-vocabulary, ootv is out-of-training-vocabulary, ooev is out-of-embedding-vocabulary.<\/span><\/span><\/span>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f0\u0001Z5a6I(o\fL\u0002~1f\u0019[<\/span><\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 Q: How CNN and LSTM work?<\/span><\/span><\/span><\/p>\n

\nE,j\u0001\\\u0015z1Z\u0013s#^\u0003I0K\u001aj\u000bp0<\/span><\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 A: The authors constructed a neural sequence labeling framework with three layers , a character sequence representation layer, a word sequence representation layer and an inference layer. LSTM and CNN are used on the character and word level, to caputure sequence representation information without hand-engineered features. More detailed information can be seen in figure 2 and 3<\/span><\/span><\/span>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f0:]3D7j\u000eB\u0004v\u001fY%Y(~(A<\/span><\/p>\n

\u00a0<\/span><\/span><\/p>\n

\na\u0014J\u0011|4p\bT\u001ah
\nj5W.E0<\/span><\/p>\n

\u00a0 \u00a0 \u00a0 \u00a0 The powerpoint of the presentation can be download.<\/span><\/span><\/span>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f07E\u001e@ L’M1b\u001eu!h<\/span><\/p>\n

\u00a0<\/span><\/span>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u4fe1\u606f\u68c0\u7d22\u5171\u4eab\u5e73\u53f0\u000fo\u0004N\/{+J4`\u0004b8v\u000fD
\nr<\/span><\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"